rpms/pacemaker
7
0
Fork 0
Code Issues Pull Requests Releases Activity

import pacemaker-2.1.2-4.el8_6.2

Browse Source
This commit is contained in:
CentOS Sources 2022-05-12 04:15:11 +00:00 committed by Stepan Oksanichenko
parent b8fffede8f
commit b82d1e1842
4 changed files with 1902 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

806
SOURCES/024-stop_unexpected.patch Normal file
View File

@ -0,0 +1,806 @@
From 767b5552ab49850204692c2c990dfb41d37589f3 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Mon, 28 Mar 2022 18:11:52 -0500
Subject: [PATCH 1/9] Refactor: libpacemaker: drop unnecessary argument from
"rsc-action" message
9875cab129 moved the setting of the "moving" variable from LogActions() to a
new "rsc-action" message, but continued to pass the variable unnecessarily
Also simplify how it's set
---
lib/pacemaker/pcmk_output.c | 10 ++++------
lib/pacemaker/pcmk_sched_native.c | 4 +---
2 files changed, 5 insertions(+), 9 deletions(-)
diff --git a/lib/pacemaker/pcmk_output.c b/lib/pacemaker/pcmk_output.c
index d864c8bd2..56963a93f 100644
--- a/lib/pacemaker/pcmk_output.c
+++ b/lib/pacemaker/pcmk_output.c
@@ -873,19 +873,18 @@ digests_xml(pcmk__output_t *out, va_list args)
} \
} while(0)
-PCMK__OUTPUT_ARGS("rsc-action", "pe_resource_t *", "pe_node_t *", "pe_node_t *",
- "gboolean")
+PCMK__OUTPUT_ARGS("rsc-action", "pe_resource_t *", "pe_node_t *", "pe_node_t *")
static int
rsc_action_default(pcmk__output_t *out, va_list args)
{
pe_resource_t *rsc = va_arg(args, pe_resource_t *);
pe_node_t *current = va_arg(args, pe_node_t *);
pe_node_t *next = va_arg(args, pe_node_t *);
- gboolean moving = va_arg(args, gboolean);
GList *possible_matches = NULL;
char *key = NULL;
int rc = pcmk_rc_no_output;
+ bool moving = false;
pe_node_t *start_node = NULL;
pe_action_t *start = NULL;
@@ -901,9 +900,8 @@ rsc_action_default(pcmk__output_t *out, va_list args)
return rc;
}
- if (current != NULL && next != NULL && !pcmk__str_eq(current->details->id, next->details->id, pcmk__str_casei)) {
- moving = TRUE;
- }
+ moving = (current != NULL) && (next != NULL)
+ && (current->details != next->details);
possible_matches = pe__resource_actions(rsc, next, RSC_START, FALSE);
if (possible_matches) {
diff --git a/lib/pacemaker/pcmk_sched_resource.c b/lib/pacemaker/pcmk_sched_resource.c
index a3d646775..41631da3d 100644
--- a/lib/pacemaker/pcmk_sched_native.c
+++ b/lib/pacemaker/pcmk_sched_native.c
@@ -2037,8 +2037,6 @@ LogActions(pe_resource_t * rsc, pe_working_set_t * data_set)
pe_node_t *next = NULL;
pe_node_t *current = NULL;
- gboolean moving = FALSE;
-
if(rsc->variant == pe_container) {
pcmk__bundle_log_actions(rsc, data_set);
return;
@@ -2066,7 +2064,7 @@ LogActions(pe_resource_t * rsc, pe_working_set_t * data_set)
return;
}
- out->message(out, "rsc-action", rsc, current, next, moving);
+ out->message(out, "rsc-action", rsc, current, next);
}
gboolean
--
2.27.0
From 870fb19715618c4ceab9ed4ae13a99658440b662 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Tue, 22 Mar 2022 15:22:23 -0500
Subject: [PATCH 2/9] Refactor: scheduler: functionize scheduling restart
actions
native_create_actions() is already overlarge, and more needs to be added to it
---
lib/pacemaker/pcmk_sched_native.c | 85 ++++++++++++++++++++-----------
1 file changed, 54 insertions(+), 31 deletions(-)
diff --git a/lib/pacemaker/pcmk_sched_native.c b/lib/pacemaker/pcmk_sched_native.c
index 808e97540..b8a1c1e1a 100644
--- a/lib/pacemaker/pcmk_sched_native.c
+++ b/lib/pacemaker/pcmk_sched_native.c
@@ -1185,6 +1185,58 @@ handle_migration_actions(pe_resource_t * rsc, pe_node_t *current, pe_node_t *cho
}
}
+/*!
+ * \internal
+ * \brief Schedule actions to bring resource down and back to current role
+ *
+ * \param[in] rsc Resource to restart
+ * \param[in] current Node that resource should be brought down on
+ * \param[in] chosen Node that resource should be brought up on
+ * \param[in] need_stop Whether the resource must be stopped
+ * \param[in] need_promote Whether the resource must be promoted
+ *
+ * \return Role that resource would have after scheduled actions are taken
+ */
+static void
+schedule_restart_actions(pe_resource_t *rsc, pe_node_t *current,
+ pe_node_t *chosen, bool need_stop, bool need_promote)
+{
+ enum rsc_role_e role = rsc->role;
+ enum rsc_role_e next_role;
+
+ // Bring resource down to a stop on its current node
+ while (role != RSC_ROLE_STOPPED) {
+ next_role = rsc_state_matrix[role][RSC_ROLE_STOPPED];
+ pe_rsc_trace(rsc, "Creating %s action to take %s down from %s to %s",
+ (need_stop? "required" : "optional"), rsc->id,
+ role2text(role), role2text(next_role));
+ if (!rsc_action_matrix[role][next_role](rsc, current, !need_stop,
+ rsc->cluster)) {
+ break;
+ }
+ role = next_role;
+ }
+
+ // Bring resource up to its next role on its next node
+ while ((rsc->role <= rsc->next_role) && (role != rsc->role)
+ && !pcmk_is_set(rsc->flags, pe_rsc_block)) {
+ bool required = need_stop;
+
+ next_role = rsc_state_matrix[role][rsc->role];
+ if ((next_role == RSC_ROLE_PROMOTED) && need_promote) {
+ required = true;
+ }
+ pe_rsc_trace(rsc, "Creating %s action to take %s up from %s to %s",
+ (required? "required" : "optional"), rsc->id,
+ role2text(role), role2text(next_role));
+ if (!rsc_action_matrix[role][next_role](rsc, chosen, !required,
+ rsc->cluster)) {
+ break;
+ }
+ role = next_role;
+ }
+}
+
void
native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
{
@@ -1332,39 +1384,10 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
/* Create any additional actions required when bringing resource down and
* back up to same level.
*/
- role = rsc->role;
- while (role != RSC_ROLE_STOPPED) {
- next_role = rsc_state_matrix[role][RSC_ROLE_STOPPED];
- pe_rsc_trace(rsc, "Creating %s action to take %s down from %s to %s",
- (need_stop? "required" : "optional"), rsc->id,
- role2text(role), role2text(next_role));
- if (rsc_action_matrix[role][next_role] (rsc, current, !need_stop, data_set) == FALSE) {
- break;
- }
- role = next_role;
- }
-
-
- while ((rsc->role <= rsc->next_role) && (role != rsc->role)
- && !pcmk_is_set(rsc->flags, pe_rsc_block)) {
- bool required = need_stop;
-
- next_role = rsc_state_matrix[role][rsc->role];
- if ((next_role == RSC_ROLE_PROMOTED) && need_promote) {
- required = true;
- }
- pe_rsc_trace(rsc, "Creating %s action to take %s up from %s to %s",
- (required? "required" : "optional"), rsc->id,
- role2text(role), role2text(next_role));
- if (rsc_action_matrix[role][next_role](rsc, chosen, !required,
- data_set) == FALSE) {
- break;
- }
- role = next_role;
- }
- role = rsc->role;
+ schedule_restart_actions(rsc, current, chosen, need_stop, need_promote);
/* Required steps from this role to the next */
+ role = rsc->role;
while (role != rsc->next_role) {
next_role = rsc_state_matrix[role][rsc->next_role];
pe_rsc_trace(rsc, "Creating action to take %s from %s to %s (ending at %s)",
--
2.27.0
From 736d4d8f5e432acf12e577d137e9165904c71b3b Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Mon, 28 Mar 2022 17:42:26 -0500
Subject: [PATCH 3/9] Log: scheduler: improve trace messages when creating
actions
---
lib/pacemaker/pcmk_sched_native.c | 22 ++++++++++++++++------
1 file changed, 16 insertions(+), 6 deletions(-)
diff --git a/lib/pacemaker/pcmk_sched_native.c b/lib/pacemaker/pcmk_sched_native.c
index b8a1c1e1a..8b651ebd2 100644
--- a/lib/pacemaker/pcmk_sched_native.c
+++ b/lib/pacemaker/pcmk_sched_native.c
@@ -1997,7 +1997,6 @@ StopRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_set
GList *gIter = NULL;
CRM_ASSERT(rsc);
- pe_rsc_trace(rsc, "%s", rsc->id);
for (gIter = rsc->running_on; gIter != NULL; gIter = gIter->next) {
pe_node_t *current = (pe_node_t *) gIter->data;
@@ -2005,16 +2004,23 @@ StopRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_set
if (rsc->partial_migration_target) {
if (rsc->partial_migration_target->details == current->details) {
- pe_rsc_trace(rsc, "Filtered %s -> %s %s", current->details->uname,
- next->details->uname, rsc->id);
+ pe_rsc_trace(rsc,
+ "Skipping stop of %s on %s "
+ "because migration to %s in progress",
+ rsc->id, current->details->uname,
+ next->details->uname);
continue;
} else {
- pe_rsc_trace(rsc, "Forced on %s %s", current->details->uname, rsc->id);
+ pe_rsc_trace(rsc,
+ "Forcing stop of %s on %s "
+ "because migration target changed",
+ rsc->id, current->details->uname);
optional = FALSE;
}
}
- pe_rsc_trace(rsc, "%s on %s", rsc->id, current->details->uname);
+ pe_rsc_trace(rsc, "Scheduling stop of %s on %s",
+ rsc->id, current->details->uname);
stop = stop_action(rsc, current, optional);
if(rsc->allocated_to == NULL) {
@@ -2048,7 +2054,11 @@ StartRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_se
pe_action_t *start = NULL;
CRM_ASSERT(rsc);
- pe_rsc_trace(rsc, "%s on %s %d %d", rsc->id, next ? next->details->uname : "N/A", optional, next ? next->weight : 0);
+
+ pe_rsc_trace(rsc, "Scheduling %s start of %s on %s (weight=%d)",
+ (optional? "optional" : "required"), rsc->id,
+ ((next == NULL)? "N/A" : next->details->uname),
+ ((next == NULL)? 0 : next->weight));
start = start_action(rsc, next, TRUE);
pcmk__order_vs_unfence(rsc, next, start, pe_order_implies_then, data_set);
--
2.27.0
From 6f987234d5246ed50f4fe2db90e5edb6a23e877d Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Tue, 1 Mar 2022 16:42:06 -0600
Subject: [PATCH 4/9] Log: scheduler: log a warning if invalid value is given
for multiple-active
---
lib/pengine/complex.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/lib/pengine/complex.c b/lib/pengine/complex.c
index e82af2aae..f2caef831 100644
--- a/lib/pengine/complex.c
+++ b/lib/pengine/complex.c
@@ -694,7 +694,12 @@ common_unpack(xmlNode * xml_obj, pe_resource_t ** rsc,
(*rsc)->recovery_type = recovery_block;
pe_rsc_trace((*rsc), "\tMultiple running resource recovery: block");
- } else {
+ } else { // "stop_start"
+ if (!pcmk__str_eq(value, "stop_start",
+ pcmk__str_casei|pcmk__str_null_matches)) {
+ pe_warn("%s is not a valid value for " XML_RSC_ATTR_MULTIPLE
+ ", using default of \"stop_start\"", value);
+ }
(*rsc)->recovery_type = recovery_stop_start;
pe_rsc_trace((*rsc), "\tMultiple running resource recovery: stop/start");
}
--
2.27.0
From 50456c3e229a6021ca0ba7346af41cd234abcc16 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Tue, 1 Mar 2022 16:49:31 -0600
Subject: [PATCH 5/9] API: libpe_status: add recovery_stop_unexpected to enum
rsc_recovery_type
The behavior is not implemented as of this commit
---
include/crm/pengine/common.h | 14 ++++++++++++--
lib/pengine/complex.c | 5 +++++
lib/pengine/native.c | 7 +++++--
3 files changed, 22 insertions(+), 4 deletions(-)
diff --git a/include/crm/pengine/common.h b/include/crm/pengine/common.h
index efe89a171..9b9f38f3b 100644
--- a/include/crm/pengine/common.h
+++ b/include/crm/pengine/common.h
@@ -1,5 +1,5 @@
/*
- * Copyright 2004-2021 the Pacemaker project contributors
+ * Copyright 2004-2022 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
@@ -78,7 +78,8 @@ enum action_tasks {
enum rsc_recovery_type {
recovery_stop_start,
recovery_stop_only,
- recovery_block
+ recovery_block,
+ recovery_stop_unexpected,
};
enum rsc_start_requirement {
@@ -143,6 +144,13 @@ const char *fail2text(enum action_fail_response fail);
const char *pe_pref(GHashTable * options, const char *name);
void calculate_active_ops(GList * sorted_op_list, int *start_index, int *stop_index);
+/*!
+ * \brief Get readable description of a recovery type
+ *
+ * \param[in] type Recovery type
+ *
+ * \return Static string describing \p type
+ */
static inline const char *
recovery2text(enum rsc_recovery_type type)
{
@@ -153,6 +161,8 @@ recovery2text(enum rsc_recovery_type type)
return "attempting recovery";
case recovery_block:
return "waiting for an administrator";
+ case recovery_stop_unexpected:
+ return "stopping unexpected instances";
}
return "Unknown";
}
diff --git a/lib/pengine/complex.c b/lib/pengine/complex.c
index f2caef831..fc9028e81 100644
--- a/lib/pengine/complex.c
+++ b/lib/pengine/complex.c
@@ -694,6 +694,11 @@ common_unpack(xmlNode * xml_obj, pe_resource_t ** rsc,
(*rsc)->recovery_type = recovery_block;
pe_rsc_trace((*rsc), "\tMultiple running resource recovery: block");
+ } else if (pcmk__str_eq(value, "stop_unexpected", pcmk__str_casei)) {
+ (*rsc)->recovery_type = recovery_stop_unexpected;
+ pe_rsc_trace((*rsc), "\tMultiple running resource recovery: "
+ "stop unexpected instances");
+
} else { // "stop_start"
if (!pcmk__str_eq(value, "stop_start",
pcmk__str_casei|pcmk__str_null_matches)) {
diff --git a/lib/pengine/native.c b/lib/pengine/native.c
index e16e54bae..fa7dc8960 100644
--- a/lib/pengine/native.c
+++ b/lib/pengine/native.c
@@ -149,8 +149,6 @@ native_add_running(pe_resource_t * rsc, pe_node_t * node, pe_working_set_t * dat
}
}
break;
- case recovery_stop_start:
- break;
case recovery_block:
pe__clear_resource_flags(rsc, pe_rsc_managed);
pe__set_resource_flags(rsc, pe_rsc_block);
@@ -171,6 +169,11 @@ native_add_running(pe_resource_t * rsc, pe_node_t * node, pe_working_set_t * dat
}
}
break;
+ default: // recovery_stop_start, recovery_stop_unexpected
+ /* The scheduler will do the right thing because the relevant
+ * variables and flags are set when unpacking the history.
+ */
+ break;
}
crm_debug("%s is active on multiple nodes including %s: %s",
rsc->id, node->details->uname,
--
2.27.0
From 5e994f0633b27e7a53701e0954466739c8f1acf7 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Wed, 30 Mar 2022 16:26:19 -0500
Subject: [PATCH 6/9] API: libpe_status: add pe_rsc_stop_unexpected flag
---
include/crm/pengine/pe_types.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/crm/pengine/pe_types.h b/include/crm/pengine/pe_types.h
index e3ecaa823..7d5394bff 100644
--- a/include/crm/pengine/pe_types.h
+++ b/include/crm/pengine/pe_types.h
@@ -277,6 +277,7 @@ struct pe_node_s {
# define pe_rsc_starting 0x00100000ULL
# define pe_rsc_stopping 0x00200000ULL
+# define pe_rsc_stop_unexpected 0x00400000ULL
# define pe_rsc_allow_migrate 0x00800000ULL
# define pe_rsc_failure_ignored 0x01000000ULL
--
2.27.0
From c1acf05be853d99c17761759b8c961f2ec4a55c2 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Thu, 31 Mar 2022 09:56:34 -0500
Subject: [PATCH 7/9] API: libpe_status: add pe_rsc_restarting flag
This is used to indicate that any actions currently being scheduled are part of
the resource's restart actions (i.e. we are in schedule_restart_actions()).
---
include/crm/pengine/pe_types.h | 1 +
lib/pacemaker/pcmk_sched_native.c | 4 ++++
2 files changed, 5 insertions(+)
diff --git a/include/crm/pengine/pe_types.h b/include/crm/pengine/pe_types.h
index 7d5394bff..77d28e900 100644
--- a/include/crm/pengine/pe_types.h
+++ b/include/crm/pengine/pe_types.h
@@ -265,6 +265,7 @@ struct pe_node_s {
# define pe_rsc_provisional 0x00000100ULL
# define pe_rsc_allocating 0x00000200ULL
# define pe_rsc_merging 0x00000400ULL
+# define pe_rsc_restarting 0x00000800ULL
# define pe_rsc_stop 0x00001000ULL
# define pe_rsc_reload 0x00002000ULL
diff --git a/lib/pacemaker/pcmk_sched_native.c b/lib/pacemaker/pcmk_sched_native.c
index 8b651ebd2..8002938b5 100644
--- a/lib/pacemaker/pcmk_sched_native.c
+++ b/lib/pacemaker/pcmk_sched_native.c
@@ -1204,6 +1204,8 @@ schedule_restart_actions(pe_resource_t *rsc, pe_node_t *current,
enum rsc_role_e role = rsc->role;
enum rsc_role_e next_role;
+ pe__set_resource_flags(rsc, pe_rsc_restarting);
+
// Bring resource down to a stop on its current node
while (role != RSC_ROLE_STOPPED) {
next_role = rsc_state_matrix[role][RSC_ROLE_STOPPED];
@@ -1235,6 +1237,8 @@ schedule_restart_actions(pe_resource_t *rsc, pe_node_t *current,
}
role = next_role;
}
+
+ pe__clear_resource_flags(rsc, pe_rsc_restarting);
}
void
--
2.27.0
From 871e2201d92520039df45062afc9120fd1fb0f30 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Tue, 1 Mar 2022 17:46:39 -0600
Subject: [PATCH 8/9] Refactor: scheduler: add expected node to primitive
variant data
Nothing uses it yet
---
include/crm/pengine/internal.h | 4 ++++
lib/pengine/native.c | 38 ++++++++++++++++++++++++++++++++++
lib/pengine/variant.h | 8 +++++--
3 files changed, 48 insertions(+), 2 deletions(-)
diff --git a/include/crm/pengine/internal.h b/include/crm/pengine/internal.h
index f949684b4..f69e6bcce 100644
--- a/include/crm/pengine/internal.h
+++ b/include/crm/pengine/internal.h
@@ -579,4 +579,8 @@ xmlNode *pe__failed_probe_for_rsc(pe_resource_t *rsc, const char *name);
const char *pe__clone_child_id(pe_resource_t *rsc);
+void pe__update_expected_node(pe_resource_t *rsc, pe_node_t *node,
+ int execution_status, int exit_status,
+ int expected_exit_status);
+
#endif
diff --git a/lib/pengine/native.c b/lib/pengine/native.c
index fa7dc8960..591d1c6f5 100644
--- a/lib/pengine/native.c
+++ b/lib/pengine/native.c
@@ -1376,3 +1376,41 @@ pe__native_is_filtered(pe_resource_t *rsc, GList *only_rsc, gboolean check_paren
return TRUE;
}
+
+/*!
+ * \internal
+ * \brief Set a resource's expected node if appropriate for a history result
+ *
+ * \param[in] rsc Resource to set expected node for
+ * \param[in] node Node to set as expected node
+ * \param[in] execution_status History entry's execution status
+ * \param[in] exit_status History entry's actual exit status
+ * \param[in] expected_status History entry's expected exit status
+ */
+void
+pe__update_expected_node(pe_resource_t *rsc, pe_node_t *node,
+ int execution_status, int exit_status,
+ int expected_exit_status)
+{
+ native_variant_data_t *native_data = NULL;
+
+ get_native_variant_data(native_data, rsc);
+
+ if ((rsc->recovery_type == recovery_stop_unexpected)
+ && (rsc->role > RSC_ROLE_STOPPED)
+ && (execution_status == PCMK_EXEC_DONE)
+ && (exit_status == expected_exit_status)) {
+ // Resource is active and was expected on this node
+ pe_rsc_trace(rsc, "Found expected node %s for %s",
+ node->details->uname, rsc->id);
+ native_data->expected_node = node;
+ pe__set_resource_flags(rsc, pe_rsc_stop_unexpected);
+
+ } else if ((native_data->expected_node != NULL)
+ && (native_data->expected_node->details == node->details)) {
+ // Resource is not cleanly active here
+ pe_rsc_trace(rsc, "Clearing expected node for %s", rsc->id);
+ native_data->expected_node = NULL;
+ pe__clear_resource_flags(rsc, pe_rsc_stop_unexpected);
+ }
+}
diff --git a/lib/pengine/variant.h b/lib/pengine/variant.h
index cabfbe81f..d8fefa9d6 100644
--- a/lib/pengine/variant.h
+++ b/lib/pengine/variant.h
@@ -1,5 +1,5 @@
/*
- * Copyright 2004-2021 the Pacemaker project contributors
+ * Copyright 2004-2022 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
@@ -139,7 +139,11 @@ typedef struct group_variant_data_s {
# elif VARIANT_NATIVE
typedef struct native_variant_data_s {
- int dummy;
+ /* If the resource is multiply active, and has multiple-active set to
+ * stop_unexpected, this will be set to the node where the resource was
+ * found active by an operation with a expected result.
+ */
+ pe_node_t *expected_node;
} native_variant_data_t;
# define get_native_variant_data(data, rsc) \
--
2.27.0
From 0e4e17e972f1c3663389f18d8f8c527bd819b3c5 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Thu, 7 Apr 2022 10:20:00 -0500
Subject: [PATCH 9/9] Feature: scheduler: implement
multiple-active=stop_unexpected
The default multiple-active policy of restarting the resource on all nodes
requires no special handling, because at least one of the locations will have
an unexpected rc, causing the resource to be marked as failed and restarted,
and StopRsc() creates stops on all nodes running the resource.
The new stop_unexpected behavior relies on most of the same handling, but
the action creation functions need to skip the node where the resource had the
expected result. For that, we set the new rsc->expected_node when unpacking a
successful result, to be checked by those functions.
Note that this still schedules a start for the resource, which is a pseudo-op
for the resource itself, but (properly) causes any dependent resources to be
restarted.
Fixes T23
---
lib/pacemaker/pcmk_output.c | 10 ++++
lib/pacemaker/pcmk_sched_native.c | 94 ++++++++++++++++++++++++++++++-
lib/pengine/unpack.c | 1 +
3 files changed, 103 insertions(+), 2 deletions(-)
diff --git a/lib/pacemaker/pcmk_output.c b/lib/pacemaker/pcmk_output.c
index 56963a93f..9a522a3e5 100644
--- a/lib/pacemaker/pcmk_output.c
+++ b/lib/pacemaker/pcmk_output.c
@@ -918,6 +918,16 @@ rsc_action_default(pcmk__output_t *out, va_list args)
if (possible_matches) {
stop = possible_matches->data;
g_list_free(possible_matches);
+ } else if (pcmk_is_set(rsc->flags, pe_rsc_stop_unexpected)) {
+ /* The resource is multiply active with multiple-active set to
+ * stop_unexpected, and not stopping on its current node, but it should
+ * be stopping elsewhere.
+ */
+ possible_matches = pe__resource_actions(rsc, NULL, RSC_STOP, FALSE);
+ if (possible_matches != NULL) {
+ stop = possible_matches->data;
+ g_list_free(possible_matches);
+ }
}
possible_matches = pe__resource_actions(rsc, next, RSC_PROMOTE, FALSE);
diff --git a/lib/pacemaker/pcmk_sched_native.c b/lib/pacemaker/pcmk_sched_native.c
index 8002938b5..c0224849f 100644
--- a/lib/pacemaker/pcmk_sched_native.c
+++ b/lib/pacemaker/pcmk_sched_native.c
@@ -1259,7 +1259,10 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
enum rsc_role_e role = RSC_ROLE_UNKNOWN;
enum rsc_role_e next_role = RSC_ROLE_UNKNOWN;
- CRM_ASSERT(rsc);
+ native_variant_data_t *native_data = NULL;
+
+ get_native_variant_data(native_data, rsc);
+
chosen = rsc->allocated_to;
next_role = rsc->next_role;
if (next_role == RSC_ROLE_UNKNOWN) {
@@ -1323,6 +1326,7 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
"(will stop on both nodes)",
rsc->id, rsc->partial_migration_source->details->uname,
rsc->partial_migration_target->details->uname);
+ multiply_active = false;
} else {
const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS);
@@ -1345,6 +1349,11 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
allow_migrate = FALSE;
}
+ if (!multiply_active) {
+ native_data->expected_node = NULL;
+ pe__clear_resource_flags(rsc, pe_rsc_stop_unexpected);
+ }
+
if (pcmk_is_set(rsc->flags, pe_rsc_start_pending)) {
pe_rsc_trace(rsc, "Creating start action for %s to represent already pending start",
rsc->id);
@@ -1995,6 +2004,32 @@ native_expand(pe_resource_t * rsc, pe_working_set_t * data_set)
out->message(out, "rsc-action", rsc, current, next);
}
+/*!
+ * \internal
+ * \brief Check whether a node is a multiply active resource's expected node
+ *
+ * \param[in] rsc Resource to check
+ * \param[in] node Node to check
+ *
+ * \return true if \p rsc is multiply active with multiple-active set to
+ * stop_unexpected, and \p node is the node where it will remain active
+ * \note This assumes that the resource's next role cannot be changed to stopped
+ * after this is called, which should be reasonable if status has already
+ * been unpacked and resources have been assigned to nodes.
+ */
+static bool
+is_expected_node(const pe_resource_t *rsc, const pe_node_t *node)
+{
+ native_variant_data_t *native_data = NULL;
+
+ get_native_variant_data(native_data, rsc);
+ return pcmk_all_flags_set(rsc->flags,
+ pe_rsc_stop_unexpected|pe_rsc_restarting)
+ && (rsc->next_role > RSC_ROLE_STOPPED)
+ && (native_data->expected_node != NULL) && (node != NULL)
+ && (native_data->expected_node->details == node->details);
+}
+
gboolean
StopRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_set_t * data_set)
{
@@ -2006,6 +2041,18 @@ StopRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_set
pe_node_t *current = (pe_node_t *) gIter->data;
pe_action_t *stop;
+ if (is_expected_node(rsc, current)) {
+ /* We are scheduling restart actions for a multiply active resource
+ * with multiple-active=stop_unexpected, and this is where it should
+ * not be stopped.
+ */
+ pe_rsc_trace(rsc,
+ "Skipping stop of multiply active resource %s "
+ "on expected node %s",
+ rsc->id, current->details->uname);
+ continue;
+ }
+
if (rsc->partial_migration_target) {
if (rsc->partial_migration_target->details == current->details) {
pe_rsc_trace(rsc,
@@ -2029,6 +2076,17 @@ StopRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_set
if(rsc->allocated_to == NULL) {
pe_action_set_reason(stop, "node availability", TRUE);
+ } else if (pcmk_is_set(rsc->flags, pe_rsc_restarting)) {
+ native_variant_data_t *native_data = NULL;
+
+ get_native_variant_data(native_data, rsc);
+ if (native_data->expected_node != NULL) {
+ /* We are stopping a multiply active resource on a node that is
+ * not its expected node, and we are still scheduling restart
+ * actions, so the stop is for being multiply active.
+ */
+ pe_action_set_reason(stop, "being multiply active", TRUE);
+ }
}
if (!pcmk_is_set(rsc->flags, pe_rsc_managed)) {
@@ -2071,6 +2129,16 @@ StartRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_se
pe__clear_action_flags(start, pe_action_optional);
}
+ if (is_expected_node(rsc, next)) {
+ /* This could be a problem if the start becomes necessary for other
+ * reasons later.
+ */
+ pe_rsc_trace(rsc,
+ "Start of multiply active resouce %s "
+ "on expected node %s will be a pseudo-action",
+ rsc->id, next->details->uname);
+ pe__set_action_flags(start, pe_action_pseudo);
+ }
return TRUE;
}
@@ -2084,6 +2152,7 @@ PromoteRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_
CRM_ASSERT(rsc);
CRM_CHECK(next != NULL, return FALSE);
+
pe_rsc_trace(rsc, "%s on %s", rsc->id, next->details->uname);
action_list = pe__resource_actions(rsc, next, RSC_START, TRUE);
@@ -2098,7 +2167,19 @@ PromoteRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_
g_list_free(action_list);
if (runnable) {
- promote_action(rsc, next, optional);
+ pe_action_t *promote = promote_action(rsc, next, optional);
+
+ if (is_expected_node(rsc, next)) {
+ /* This could be a problem if the promote becomes necessary for
+ * other reasons later.
+ */
+ pe_rsc_trace(rsc,
+ "Promotion of multiply active resouce %s "
+ "on expected node %s will be a pseudo-action",
+ rsc->id, next->details->uname);
+ pe__set_action_flags(promote, pe_action_pseudo);
+ }
+
return TRUE;
}
@@ -2122,6 +2203,15 @@ DemoteRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_s
GList *gIter = NULL;
CRM_ASSERT(rsc);
+
+ if (is_expected_node(rsc, next)) {
+ pe_rsc_trace(rsc,
+ "Skipping demote of multiply active resource %s "
+ "on expected node %s",
+ rsc->id, next->details->uname);
+ return TRUE;
+ }
+
pe_rsc_trace(rsc, "%s", rsc->id);
/* CRM_CHECK(rsc->next_role == RSC_ROLE_UNPROMOTED, return FALSE); */
diff --git a/lib/pengine/unpack.c b/lib/pengine/unpack.c
index d218f523f..edaa9de48 100644
--- a/lib/pengine/unpack.c
+++ b/lib/pengine/unpack.c
@@ -3974,6 +3974,7 @@ unpack_rsc_op(pe_resource_t *rsc, pe_node_t *node, xmlNode *xml_op,
}
done:
+ pe__update_expected_node(rsc, node, status, rc, target_rc);
pe_rsc_trace(rsc, "Resource %s after %s: role=%s, next=%s",
rsc->id, task, role2text(rsc->role),
role2text(rsc->next_role));
--
2.27.0

495
SOURCES/025-stop_unexpected-test.patch Normal file
View File

@ -0,0 +1,495 @@
From 8a0a16c8ed72c74d656664694ebe36b76ff22498 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Wed, 30 Mar 2022 17:14:33 -0500
Subject: [PATCH] Test: cts-scheduler: add test for
multiple-active=stop_unexpected
---
cts/cts-scheduler.in | 1 +
cts/scheduler/dot/stop-unexpected.dot | 40 ++++
cts/scheduler/exp/stop-unexpected.exp | 201 ++++++++++++++++++
cts/scheduler/scores/stop-unexpected.scores | 17 ++
cts/scheduler/summary/stop-unexpected.summary | 41 ++++
cts/scheduler/xml/stop-unexpected.xml | 131 ++++++++++++
6 files changed, 431 insertions(+)
create mode 100644 cts/scheduler/dot/stop-unexpected.dot
create mode 100644 cts/scheduler/exp/stop-unexpected.exp
create mode 100644 cts/scheduler/scores/stop-unexpected.scores
create mode 100644 cts/scheduler/summary/stop-unexpected.summary
create mode 100644 cts/scheduler/xml/stop-unexpected.xml
diff --git a/cts/cts-scheduler.in b/cts/cts-scheduler.in
index 3a8aeaca5..8c04687da 100644
--- a/cts/cts-scheduler.in
+++ b/cts/cts-scheduler.in
@@ -273,6 +273,7 @@ TESTS = [
[ "rec-rsc-6", "Resource Recover - multiple - restart" ],
[ "rec-rsc-7", "Resource Recover - multiple - stop" ],
[ "rec-rsc-8", "Resource Recover - multiple - block" ],
+ [ "stop-unexpected", "Resource Recover - multiple - stop unexpected" ],
[ "rec-rsc-9", "Resource Recover - group/group" ],
[ "monitor-recovery", "on-fail=block + resource recovery detected by recurring monitor" ],
[ "stop-failure-no-quorum", "Stop failure without quorum" ],
diff --git a/cts/scheduler/dot/stop-unexpected.dot b/cts/scheduler/dot/stop-unexpected.dot
new file mode 100644
index 000000000..0f67eec54
--- /dev/null
+++ b/cts/scheduler/dot/stop-unexpected.dot
@@ -0,0 +1,40 @@
+ digraph "g" {
+"dgroup_running_0" [ style=bold color="green" fontcolor="orange"]
+"dgroup_start_0" -> "dgroup_running_0" [ style = bold]
+"dgroup_start_0" -> "dummy2_start_0 node2" [ style = bold]
+"dgroup_start_0" -> "dummy3_start_0 node2" [ style = bold]
+"dgroup_start_0" -> "dummy_start_0 node2" [ style = bold]
+"dgroup_start_0" [ style=bold color="green" fontcolor="orange"]
+"dgroup_stop_0" -> "dgroup_stopped_0" [ style = bold]
+"dgroup_stop_0" -> "dummy2_stop_0 node2" [ style = bold]
+"dgroup_stop_0" -> "dummy3_stop_0 node2" [ style = bold]
+"dgroup_stop_0" -> "dummy_stop_0 node3" [ style = bold]
+"dgroup_stop_0" [ style=bold color="green" fontcolor="orange"]
+"dgroup_stopped_0" -> "dgroup_start_0" [ style = bold]
+"dgroup_stopped_0" [ style=bold color="green" fontcolor="orange"]
+"dummy2_monitor_10000 node2" [ style=bold color="green" fontcolor="black"]
+"dummy2_start_0 node2" -> "dgroup_running_0" [ style = bold]
+"dummy2_start_0 node2" -> "dummy2_monitor_10000 node2" [ style = bold]
+"dummy2_start_0 node2" -> "dummy3_start_0 node2" [ style = bold]
+"dummy2_start_0 node2" [ style=bold color="green" fontcolor="black"]
+"dummy2_stop_0 node2" -> "dgroup_stopped_0" [ style = bold]
+"dummy2_stop_0 node2" -> "dummy2_start_0 node2" [ style = bold]
+"dummy2_stop_0 node2" -> "dummy_stop_0 node3" [ style = bold]
+"dummy2_stop_0 node2" [ style=bold color="green" fontcolor="black"]
+"dummy3_monitor_10000 node2" [ style=bold color="green" fontcolor="black"]
+"dummy3_start_0 node2" -> "dgroup_running_0" [ style = bold]
+"dummy3_start_0 node2" -> "dummy3_monitor_10000 node2" [ style = bold]
+"dummy3_start_0 node2" [ style=bold color="green" fontcolor="black"]
+"dummy3_stop_0 node2" -> "dgroup_stopped_0" [ style = bold]
+"dummy3_stop_0 node2" -> "dummy2_stop_0 node2" [ style = bold]
+"dummy3_stop_0 node2" -> "dummy3_start_0 node2" [ style = bold]
+"dummy3_stop_0 node2" [ style=bold color="green" fontcolor="black"]
+"dummy_monitor_10000 node2" [ style=bold color="green" fontcolor="black"]
+"dummy_start_0 node2" -> "dgroup_running_0" [ style = bold]
+"dummy_start_0 node2" -> "dummy2_start_0 node2" [ style = bold]
+"dummy_start_0 node2" -> "dummy_monitor_10000 node2" [ style = bold]
+"dummy_start_0 node2" [ style=bold color="green" fontcolor="orange"]
+"dummy_stop_0 node3" -> "dgroup_stopped_0" [ style = bold]
+"dummy_stop_0 node3" -> "dummy_start_0 node2" [ style = bold]
+"dummy_stop_0 node3" [ style=bold color="green" fontcolor="black"]
+}
diff --git a/cts/scheduler/exp/stop-unexpected.exp b/cts/scheduler/exp/stop-unexpected.exp
new file mode 100644
index 000000000..1f94532f7
--- /dev/null
+++ b/cts/scheduler/exp/stop-unexpected.exp
@@ -0,0 +1,201 @@
+<transition_graph cluster-delay="60s" stonith-timeout="90" failed-stop-offset="INFINITY" failed-start-offset="INFINITY" transition_id="0">
+ <synapse id="0">
+ <action_set>
+ <pseudo_event id="15" operation="stopped" operation_key="dgroup_stopped_0">
+ <attributes CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </pseudo_event>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="4" operation="stop" operation_key="dummy_stop_0" on_node="node3" on_node_uuid="node3"/>
+ </trigger>
+ <trigger>
+ <rsc_op id="8" operation="stop" operation_key="dummy2_stop_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <rsc_op id="10" operation="stop" operation_key="dummy3_stop_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="14" operation="stop" operation_key="dgroup_stop_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="1">
+ <action_set>
+ <pseudo_event id="14" operation="stop" operation_key="dgroup_stop_0">
+ <attributes CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </pseudo_event>
+ </action_set>
+ <inputs/>
+ </synapse>
+ <synapse id="2">
+ <action_set>
+ <pseudo_event id="13" operation="running" operation_key="dgroup_running_0">
+ <attributes CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </pseudo_event>
+ </action_set>
+ <inputs>
+ <trigger>
+ <pseudo_event id="7" operation="start" operation_key="dummy_start_0"/>
+ </trigger>
+ <trigger>
+ <rsc_op id="9" operation="start" operation_key="dummy2_start_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <rsc_op id="11" operation="start" operation_key="dummy3_start_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="12" operation="start" operation_key="dgroup_start_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="3">
+ <action_set>
+ <pseudo_event id="12" operation="start" operation_key="dgroup_start_0">
+ <attributes CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </pseudo_event>
+ </action_set>
+ <inputs>
+ <trigger>
+ <pseudo_event id="15" operation="stopped" operation_key="dgroup_stopped_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="4">
+ <action_set>
+ <pseudo_event id="7" operation="start" operation_key="dummy_start_0">
+ <attributes CRM_meta_name="start" CRM_meta_record_pending="false" CRM_meta_timeout="300000" />
+ </pseudo_event>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="4" operation="stop" operation_key="dummy_stop_0" on_node="node3" on_node_uuid="node3"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="12" operation="start" operation_key="dgroup_start_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="5">
+ <action_set>
+ <rsc_op id="4" operation="stop" operation_key="dummy_stop_0" on_node="node3" on_node_uuid="node3">
+ <primitive id="dummy" class="ocf" provider="heartbeat" type="DummyTimeout"/>
+ <attributes CRM_meta_name="stop" CRM_meta_on_node="node3" CRM_meta_on_node_uuid="node3" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="8" operation="stop" operation_key="dummy2_stop_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="14" operation="stop" operation_key="dgroup_stop_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="6">
+ <action_set>
+ <rsc_op id="2" operation="monitor" operation_key="dummy_monitor_10000" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy" class="ocf" provider="heartbeat" type="DummyTimeout"/>
+ <attributes CRM_meta_interval="10000" CRM_meta_name="monitor" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <pseudo_event id="7" operation="start" operation_key="dummy_start_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="7">
+ <action_set>
+ <rsc_op id="9" operation="start" operation_key="dummy2_start_0" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy2" class="ocf" provider="heartbeat" type="Dummy"/>
+ <attributes CRM_meta_name="start" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <pseudo_event id="7" operation="start" operation_key="dummy_start_0"/>
+ </trigger>
+ <trigger>
+ <rsc_op id="8" operation="stop" operation_key="dummy2_stop_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="12" operation="start" operation_key="dgroup_start_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="8">
+ <action_set>
+ <rsc_op id="8" operation="stop" operation_key="dummy2_stop_0" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy2" class="ocf" provider="heartbeat" type="Dummy"/>
+ <attributes CRM_meta_name="stop" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="10" operation="stop" operation_key="dummy3_stop_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="14" operation="stop" operation_key="dgroup_stop_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="9">
+ <action_set>
+ <rsc_op id="3" operation="monitor" operation_key="dummy2_monitor_10000" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy2" class="ocf" provider="heartbeat" type="Dummy"/>
+ <attributes CRM_meta_interval="10000" CRM_meta_name="monitor" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="9" operation="start" operation_key="dummy2_start_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="10">
+ <action_set>
+ <rsc_op id="11" operation="start" operation_key="dummy3_start_0" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy3" class="ocf" provider="heartbeat" type="Dummy"/>
+ <attributes CRM_meta_name="start" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="9" operation="start" operation_key="dummy2_start_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <rsc_op id="10" operation="stop" operation_key="dummy3_stop_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ <trigger>
+ <pseudo_event id="12" operation="start" operation_key="dgroup_start_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="11">
+ <action_set>
+ <rsc_op id="10" operation="stop" operation_key="dummy3_stop_0" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy3" class="ocf" provider="heartbeat" type="Dummy"/>
+ <attributes CRM_meta_name="stop" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <pseudo_event id="14" operation="stop" operation_key="dgroup_stop_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="12">
+ <action_set>
+ <rsc_op id="1" operation="monitor" operation_key="dummy3_monitor_10000" on_node="node2" on_node_uuid="node2">
+ <primitive id="dummy3" class="ocf" provider="heartbeat" type="Dummy"/>
+ <attributes CRM_meta_interval="10000" CRM_meta_name="monitor" CRM_meta_on_node="node2" CRM_meta_on_node_uuid="node2" CRM_meta_record_pending="false" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="11" operation="start" operation_key="dummy3_start_0" on_node="node2" on_node_uuid="node2"/>
+ </trigger>
+ </inputs>
+ </synapse>
+</transition_graph>
diff --git a/cts/scheduler/scores/stop-unexpected.scores b/cts/scheduler/scores/stop-unexpected.scores
new file mode 100644
index 000000000..68f98e837
--- /dev/null
+++ b/cts/scheduler/scores/stop-unexpected.scores
@@ -0,0 +1,17 @@
+
+pcmk__group_allocate: dgroup allocation score on node2: 0
+pcmk__group_allocate: dgroup allocation score on node3: 0
+pcmk__group_allocate: dummy allocation score on node2: 0
+pcmk__group_allocate: dummy allocation score on node3: 0
+pcmk__group_allocate: dummy2 allocation score on node2: 100
+pcmk__group_allocate: dummy2 allocation score on node3: 0
+pcmk__group_allocate: dummy3 allocation score on node2: 100
+pcmk__group_allocate: dummy3 allocation score on node3: 0
+pcmk__native_allocate: dummy allocation score on node2: 200
+pcmk__native_allocate: dummy allocation score on node3: 0
+pcmk__native_allocate: dummy2 allocation score on node2: 200
+pcmk__native_allocate: dummy2 allocation score on node3: -INFINITY
+pcmk__native_allocate: dummy3 allocation score on node2: 100
+pcmk__native_allocate: dummy3 allocation score on node3: -INFINITY
+pcmk__native_allocate: st-sbd allocation score on node2: 100
+pcmk__native_allocate: st-sbd allocation score on node3: 0
diff --git a/cts/scheduler/summary/stop-unexpected.summary b/cts/scheduler/summary/stop-unexpected.summary
new file mode 100644
index 000000000..7c7fc68b6
--- /dev/null
+++ b/cts/scheduler/summary/stop-unexpected.summary
@@ -0,0 +1,41 @@
+Current cluster status:
+ * Node List:
+ * Online: [ node2 node3 ]
+
+ * Full List of Resources:
+ * st-sbd (stonith:external/sbd): Started node2
+ * Resource Group: dgroup:
+ * dummy (ocf:heartbeat:DummyTimeout): FAILED [ node2 node3 ]
+ * dummy2 (ocf:heartbeat:Dummy): Started node2
+ * dummy3 (ocf:heartbeat:Dummy): Started node2
+
+Transition Summary:
+ * Recover dummy ( node2 ) due to being multiply active
+ * Restart dummy2 ( node2 ) due to required dummy start
+ * Restart dummy3 ( node2 ) due to required dummy2 start
+
+Executing Cluster Transition:
+ * Pseudo action: dgroup_stop_0
+ * Resource action: dummy3 stop on node2
+ * Resource action: dummy2 stop on node2
+ * Resource action: dummy stop on node3
+ * Pseudo action: dgroup_stopped_0
+ * Pseudo action: dgroup_start_0
+ * Pseudo action: dummy_start_0
+ * Resource action: dummy monitor=10000 on node2
+ * Resource action: dummy2 start on node2
+ * Resource action: dummy2 monitor=10000 on node2
+ * Resource action: dummy3 start on node2
+ * Resource action: dummy3 monitor=10000 on node2
+ * Pseudo action: dgroup_running_0
+
+Revised Cluster Status:
+ * Node List:
+ * Online: [ node2 node3 ]
+
+ * Full List of Resources:
+ * st-sbd (stonith:external/sbd): Started node2
+ * Resource Group: dgroup:
+ * dummy (ocf:heartbeat:DummyTimeout): Started node2
+ * dummy2 (ocf:heartbeat:Dummy): Started node2
+ * dummy3 (ocf:heartbeat:Dummy): Started node2
diff --git a/cts/scheduler/xml/stop-unexpected.xml b/cts/scheduler/xml/stop-unexpected.xml
new file mode 100644
index 000000000..6e61aeba3
--- /dev/null
+++ b/cts/scheduler/xml/stop-unexpected.xml
@@ -0,0 +1,131 @@
+<cib epoch="631" num_updates="25" admin_epoch="0" validate-with="pacemaker-3.0" crm_feature_set="3.0.8" have-quorum="1" cib-last-written="Thu Aug 20 11:44:27 2015" update-origin="node2" update-client="cibadmin" update-user="root" dc-uuid="node2">
+ <configuration>
+ <crm_config>
+ <cluster_property_set id="cib-bootstrap-options">
+ <nvpair name="dc-version" value="1.1.11-3ca8c3b" id="cib-bootstrap-options-dc-version"/>
+ <nvpair name="cluster-infrastructure" value="corosync" id="cib-bootstrap-options-cluster-infrastructure"/>
+ <nvpair name="node-action-limit" value="2" id="cib-bootstrap-options-node-action-limit"/>
+ <nvpair name="no-quorum-policy" value="ignore" id="cib-bootstrap-options-no-quorum-policy"/>
+ <nvpair name="stonith-enabled" value="false" id="cib-bootstrap-options-stonith-enabled"/>
+ <nvpair name="stonith-timeout" value="90" id="cib-bootstrap-options-stonith-timeout"/>
+ <nvpair name="last-lrm-refresh" value="1439556204" id="cib-bootstrap-options-last-lrm-refresh"/>
+ </cluster_property_set>
+ </crm_config>
+ <nodes>
+ <node uname="node2" id="node2">
+ <instance_attributes id="nodes-node2">
+ <nvpair id="nodes-node2-standby" name="standby" value="off"/>
+ </instance_attributes>
+ </node>
+ <node id="node3" uname="node3">
+ <instance_attributes id="nodes-node3">
+ <nvpair id="nodes-node3-standby" name="standby" value="off"/>
+ </instance_attributes>
+ </node>
+ </nodes>
+ <resources>
+ <primitive id="st-sbd" class="stonith" type="external/sbd"/>
+ <group id="dgroup">
+ <meta_attributes id="dgroup-meta_attributes">
+ <nvpair name="multiple-active" value="stop_unexpected" id="dgroup-meta_attributes-multiple-active"/>
+ </meta_attributes>
+ <primitive id="dummy" class="ocf" provider="heartbeat" type="DummyTimeout">
+ <operations>
+ <op name="monitor" interval="10s" timeout="20" id="dummy-monitor-10s"/>
+ <op name="start" timeout="300s" interval="0" id="dummy-start-0"/>
+ <op name="stop" timeout="20s" interval="0" id="dummy-stop-0"/>
+ </operations>
+ <meta_attributes id="dummy-meta_attributes">
+ <nvpair name="target-role" value="Started" id="dummy-meta_attributes-target-role"/>
+ </meta_attributes>
+ </primitive>
+ <primitive id="dummy2" class="ocf" provider="heartbeat" type="Dummy">
+ <operations>
+ <op name="monitor" interval="10s" timeout="20" id="dummy2-monitor-10s"/>
+ <op name="start" timeout="20s" interval="0" id="dummy2-start-0"/>
+ <op name="stop" timeout="20s" interval="0" id="dummy2-stop-0"/>
+ </operations>
+ <meta_attributes id="dummy2-meta_attributes">
+ <nvpair name="target-role" value="Started" id="dummy2-meta_attributes-target-role"/>
+ </meta_attributes>
+ </primitive>
+ <primitive id="dummy3" class="ocf" provider="heartbeat" type="Dummy">
+ <operations>
+ <op name="monitor" interval="10s" timeout="20" id="dummy3-monitor-10s"/>
+ <op name="start" timeout="20s" interval="0" id="dummy3-start-0"/>
+ <op name="stop" timeout="20s" interval="0" id="dummy3-stop-0"/>
+ </operations>
+ <meta_attributes id="dummy3-meta_attributes">
+ <nvpair name="target-role" value="Started" id="dummy3-meta_attributes-target-role"/>
+ </meta_attributes>
+ </primitive>
+ </group>
+ </resources>
+ <constraints/>
+ <op_defaults>
+ <meta_attributes id="op_defaults-options">
+ <nvpair id="op_defaults-options-record-pending" name="record-pending" value="false"/>
+ </meta_attributes>
+ </op_defaults>
+ <rsc_defaults>
+ <meta_attributes id="rsc_defaults-options">
+ <nvpair name="resource-stickiness" value="100" id="rsc_defaults-options-resource-stickiness"/>
+ <nvpair name="migration-threshold" value="100" id="rsc_defaults-options-migration-threshold"/>
+ </meta_attributes>
+ </rsc_defaults>
+ </configuration>
+ <status>
+ <node_state id="node2" uname="node2" in_ccm="true" crmd="online" crm-debug-origin="do_update_resource" join="member" expected="member">
+ <transient_attributes id="node2">
+ <instance_attributes id="status-node2">
+ <nvpair id="status-node2-shutdown" name="shutdown" value="0"/>
+ <nvpair id="status-node2-probe_complete" name="probe_complete" value="true"/>
+ </instance_attributes>
+ </transient_attributes>
+ <lrm id="node2">
+ <lrm_resources>
+ <lrm_resource id="dummy3" type="Dummy" class="ocf" provider="heartbeat">
+ <lrm_rsc_op id="dummy3_last_0" operation_key="dummy3_start_0" operation="start" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="13:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;13:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="24" rc-code="0" op-status="0" interval="0" last-run="1440063239" last-rc-change="1440063239" exec-time="6" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ <lrm_rsc_op id="dummy3_monitor_10000" operation_key="dummy3_monitor_10000" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="14:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;14:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="25" rc-code="0" op-status="0" interval="10000" last-rc-change="1440063239" exec-time="5" queue-time="0" op-digest="4811cef7f7f94e3a35a70be7916cb2fd"/>
+ </lrm_resource>
+ <lrm_resource id="st-sbd" type="external/sbd" class="stonith">
+ <lrm_rsc_op id="st-sbd_last_0" operation_key="st-sbd_start_0" operation="start" crm-debug-origin="do_update_resource" crm_feature_set="3.0.8" transition-key="10:6:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;10:6:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="27" rc-code="0" op-status="0" interval="0" last-run="1440064019" last-rc-change="1440064019" exec-time="1213" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ <lrm_resource id="dummy" type="DummyTimeout" class="ocf" provider="heartbeat">
+ <lrm_rsc_op id="dummy_last_0" operation_key="dummy_start_0" operation="start" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="9:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;9:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="20" rc-code="0" op-status="0" interval="0" last-run="1440063237" last-rc-change="1440063237" exec-time="1009" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ <lrm_rsc_op id="dummy_monitor_10000" operation_key="dummy_monitor_10000" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="10:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;10:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="21" rc-code="0" op-status="0" interval="10000" last-rc-change="1440063238" exec-time="1010" queue-time="0" op-digest="4811cef7f7f94e3a35a70be7916cb2fd"/>
+ </lrm_resource>
+ <lrm_resource id="dummy2" type="Dummy" class="ocf" provider="heartbeat">
+ <lrm_rsc_op id="dummy2_last_0" operation_key="dummy2_start_0" operation="start" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="11:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;11:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="22" rc-code="0" op-status="0" interval="0" last-run="1440063239" last-rc-change="1440063239" exec-time="5" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ <lrm_rsc_op id="dummy2_monitor_10000" operation_key="dummy2_monitor_10000" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="12:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;12:1:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="23" rc-code="0" op-status="0" interval="10000" last-rc-change="1440063239" exec-time="5" queue-time="0" op-digest="4811cef7f7f94e3a35a70be7916cb2fd"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ <node_state id="node3" uname="node3" crmd="online" crm-debug-origin="do_update_resource" in_ccm="true" join="member" expected="member">
+ <transient_attributes id="node3">
+ <instance_attributes id="status-node3">
+ <nvpair id="status-node3-shutdown" name="shutdown" value="0"/>
+ <nvpair id="status-node3-probe_complete" name="probe_complete" value="true"/>
+ </instance_attributes>
+ </transient_attributes>
+ <lrm id="node3">
+ <lrm_resources>
+ <lrm_resource id="dummy3" type="Dummy" class="ocf" provider="heartbeat">
+ <lrm_rsc_op id="dummy3_last_0" operation_key="dummy3_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="11:2:7:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:7;11:2:7:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="18" rc-code="7" op-status="0" interval="0" last-run="1440063820" last-rc-change="1440063820" exec-time="7" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ <lrm_resource id="st-sbd" type="external/sbd" class="stonith">
+ <lrm_rsc_op id="st-sbd_last_0" operation_key="st-sbd_stop_0" operation="stop" crm-debug-origin="do_update_resource" crm_feature_set="3.0.8" transition-key="9:6:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:0;9:6:0:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="21" rc-code="0" op-status="0" interval="0" last-run="1440064019" last-rc-change="1440064019" exec-time="1" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ <lrm_resource id="dummy" type="DummyTimeout" class="ocf" provider="heartbeat">
+ <lrm_rsc_op id="dummy_last_0" operation_key="dummy_start_0" operation="start" crm-debug-origin="do_update_resource" crm_feature_set="3.0.8" transition-key="5:0:0:a5e85e43-f35a-4f75-8e15-f0ddc8d81812" transition-magic="0:7;5:0:0:a5e85e43-f35a-4f75-8e15-f0ddc8d81812" call-id="20" rc-code="7" op-status="0" interval="0" last-run="1440063984" last-rc-change="1440063984" exec-time="179014" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ <lrm_rsc_op id="dummy_last_failure_0" operation_key="dummy_start_0" operation="start" crm-debug-origin="do_update_resource" crm_feature_set="3.0.8" transition-key="5:0:0:a5e85e43-f35a-4f75-8e15-f0ddc8d81812" transition-magic="0:7;5:0:0:a5e85e43-f35a-4f75-8e15-f0ddc8d81812" call-id="20" rc-code="7" op-status="0" interval="0" last-run="1440063984" last-rc-change="1440063984" exec-time="179014" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ <lrm_resource id="dummy2" type="Dummy" class="ocf" provider="heartbeat">
+ <lrm_rsc_op id="dummy2_last_0" operation_key="dummy2_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.0.8" transition-key="10:2:7:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" transition-magic="0:7;10:2:7:b78eae54-472e-4e90-a3c5-ec4b25a6d8cf" call-id="14" rc-code="7" op-status="0" interval="0" last-run="1440063820" last-rc-change="1440063820" exec-time="11" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ </status>
+</cib>
--
2.27.0

589
SOURCES/026-stop_unexpected-fix.patch Normal file
View File

@ -0,0 +1,589 @@
From 4a5dcc5210160f7d167bc68142635c1b5a6d4af2 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Fri, 22 Apr 2022 10:47:29 -0500
Subject: [PATCH 1/3] Fix: scheduler: make multiple-active="stop_unexpected"
actually work
The previous implementation covered the scenario in the regression test and not
much else. It unnecessarily added an expected_node member to the native variant
data, when the resource's allocated_to is sufficient to know the expected node.
---
lib/pacemaker/pcmk_sched_native.c | 45 +++++++++++++++----------------
lib/pengine/unpack.c | 1 -
2 files changed, 22 insertions(+), 24 deletions(-)
diff --git a/lib/pacemaker/pcmk_sched_native.c b/lib/pacemaker/pcmk_sched_native.c
index c0224849f..a1a51721e 100644
--- a/lib/pacemaker/pcmk_sched_native.c
+++ b/lib/pacemaker/pcmk_sched_native.c
@@ -1250,7 +1250,7 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
gboolean need_stop = FALSE;
bool need_promote = FALSE;
gboolean is_moving = FALSE;
- gboolean allow_migrate = pcmk_is_set(rsc->flags, pe_rsc_allow_migrate)? TRUE : FALSE;
+ gboolean allow_migrate = FALSE;
GList *gIter = NULL;
unsigned int num_all_active = 0;
@@ -1259,9 +1259,8 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
enum rsc_role_e role = RSC_ROLE_UNKNOWN;
enum rsc_role_e next_role = RSC_ROLE_UNKNOWN;
- native_variant_data_t *native_data = NULL;
-
- get_native_variant_data(native_data, rsc);
+ CRM_ASSERT(rsc != NULL);
+ allow_migrate = pcmk_is_set(rsc->flags, pe_rsc_allow_migrate)? TRUE : FALSE;
chosen = rsc->allocated_to;
next_role = rsc->next_role;
@@ -1338,8 +1337,16 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
crm_notice("See https://wiki.clusterlabs.org/wiki/FAQ#Resource_is_Too_Active for more information");
}
- if (rsc->recovery_type == recovery_stop_start) {
- need_stop = TRUE;
+ switch (rsc->recovery_type) {
+ case recovery_stop_start:
+ need_stop = TRUE;
+ break;
+ case recovery_stop_unexpected:
+ need_stop = TRUE; // StopRsc() will skip expected node
+ pe__set_resource_flags(rsc, pe_rsc_stop_unexpected);
+ break;
+ default:
+ break;
}
/* If by chance a partial migration is in process, but the migration
@@ -1350,7 +1357,6 @@ native_create_actions(pe_resource_t * rsc, pe_working_set_t * data_set)
}
if (!multiply_active) {
- native_data->expected_node = NULL;
pe__clear_resource_flags(rsc, pe_rsc_stop_unexpected);
}
@@ -2020,14 +2026,11 @@ native_expand(pe_resource_t * rsc, pe_working_set_t * data_set)
static bool
is_expected_node(const pe_resource_t *rsc, const pe_node_t *node)
{
- native_variant_data_t *native_data = NULL;
-
- get_native_variant_data(native_data, rsc);
return pcmk_all_flags_set(rsc->flags,
pe_rsc_stop_unexpected|pe_rsc_restarting)
&& (rsc->next_role > RSC_ROLE_STOPPED)
- && (native_data->expected_node != NULL) && (node != NULL)
- && (native_data->expected_node->details == node->details);
+ && (rsc->allocated_to != NULL) && (node != NULL)
+ && (rsc->allocated_to->details == node->details);
}
gboolean
@@ -2076,17 +2079,13 @@ StopRsc(pe_resource_t * rsc, pe_node_t * next, gboolean optional, pe_working_set
if(rsc->allocated_to == NULL) {
pe_action_set_reason(stop, "node availability", TRUE);
- } else if (pcmk_is_set(rsc->flags, pe_rsc_restarting)) {
- native_variant_data_t *native_data = NULL;
-
- get_native_variant_data(native_data, rsc);
- if (native_data->expected_node != NULL) {
- /* We are stopping a multiply active resource on a node that is
- * not its expected node, and we are still scheduling restart
- * actions, so the stop is for being multiply active.
- */
- pe_action_set_reason(stop, "being multiply active", TRUE);
- }
+ } else if (pcmk_all_flags_set(rsc->flags, pe_rsc_restarting
+ |pe_rsc_stop_unexpected)) {
+ /* We are stopping a multiply active resource on a node that is
+ * not its expected node, and we are still scheduling restart
+ * actions, so the stop is for being multiply active.
+ */
+ pe_action_set_reason(stop, "being multiply active", TRUE);
}
if (!pcmk_is_set(rsc->flags, pe_rsc_managed)) {
diff --git a/lib/pengine/unpack.c b/lib/pengine/unpack.c
index 17dea0d7a..426022013 100644
--- a/lib/pengine/unpack.c
+++ b/lib/pengine/unpack.c
@@ -3945,7 +3945,6 @@ unpack_rsc_op(pe_resource_t *rsc, pe_node_t *node, xmlNode *xml_op,
}
done:
- pe__update_expected_node(rsc, node, status, rc, target_rc);
pe_rsc_trace(rsc, "Resource %s after %s: role=%s, next=%s",
rsc->id, task, role2text(rsc->role),
role2text(rsc->next_role));
--
2.27.0
From 703d3a09bce389afb4e095e1ac7af29eb5edd189 Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Fri, 22 Apr 2022 14:02:34 -0500
Subject: [PATCH 2/3] Test: scheduler: add a second regression test for
multiple-active=stop_unexpected
---
cts/cts-scheduler.in | 3 +-
cts/scheduler/dot/stop-unexpected-2.dot | 7 +
cts/scheduler/exp/stop-unexpected-2.exp | 36 ++++
cts/scheduler/scores/stop-unexpected-2.scores | 21 ++
.../summary/stop-unexpected-2.summary | 29 +++
cts/scheduler/xml/stop-unexpected-2.xml | 204 ++++++++++++++++++
6 files changed, 299 insertions(+), 1 deletion(-)
create mode 100644 cts/scheduler/dot/stop-unexpected-2.dot
create mode 100644 cts/scheduler/exp/stop-unexpected-2.exp
create mode 100644 cts/scheduler/scores/stop-unexpected-2.scores
create mode 100644 cts/scheduler/summary/stop-unexpected-2.summary
create mode 100644 cts/scheduler/xml/stop-unexpected-2.xml
diff --git a/cts/cts-scheduler.in b/cts/cts-scheduler.in
index 8c04687da..7fc76cce4 100644
--- a/cts/cts-scheduler.in
+++ b/cts/cts-scheduler.in
@@ -273,8 +273,9 @@ TESTS = [
[ "rec-rsc-6", "Resource Recover - multiple - restart" ],
[ "rec-rsc-7", "Resource Recover - multiple - stop" ],
[ "rec-rsc-8", "Resource Recover - multiple - block" ],
- [ "stop-unexpected", "Resource Recover - multiple - stop unexpected" ],
[ "rec-rsc-9", "Resource Recover - group/group" ],
+ [ "stop-unexpected", "Recover multiply active group with stop_unexpected" ],
+ [ "stop-unexpected-2", "Resource multiply active primitve with stop_unexpected" ],
[ "monitor-recovery", "on-fail=block + resource recovery detected by recurring monitor" ],
[ "stop-failure-no-quorum", "Stop failure without quorum" ],
[ "stop-failure-no-fencing", "Stop failure without fencing available" ],
diff --git a/cts/scheduler/dot/stop-unexpected-2.dot b/cts/scheduler/dot/stop-unexpected-2.dot
new file mode 100644
index 000000000..cdaebf551
--- /dev/null
+++ b/cts/scheduler/dot/stop-unexpected-2.dot
@@ -0,0 +1,7 @@
+ digraph "g" {
+"test_monitor_10000 rhel8-4" [ style=bold color="green" fontcolor="black"]
+"test_start_0 rhel8-4" -> "test_monitor_10000 rhel8-4" [ style = bold]
+"test_start_0 rhel8-4" [ style=bold color="green" fontcolor="orange"]
+"test_stop_0 rhel8-3" -> "test_start_0 rhel8-4" [ style = bold]
+"test_stop_0 rhel8-3" [ style=bold color="green" fontcolor="black"]
+}
diff --git a/cts/scheduler/exp/stop-unexpected-2.exp b/cts/scheduler/exp/stop-unexpected-2.exp
new file mode 100644
index 000000000..258053c08
--- /dev/null
+++ b/cts/scheduler/exp/stop-unexpected-2.exp
@@ -0,0 +1,36 @@
+<transition_graph cluster-delay="60s" stonith-timeout="60s" failed-stop-offset="INFINITY" failed-start-offset="1" transition_id="0">
+ <synapse id="0">
+ <action_set>
+ <rsc_op id="10" operation="monitor" operation_key="test_monitor_10000" on_node="rhel8-4" on_node_uuid="4">
+ <primitive id="test" class="ocf" provider="pacemaker" type="Dummy"/>
+ <attributes CRM_meta_interval="10000" CRM_meta_name="monitor" CRM_meta_on_node="rhel8-4" CRM_meta_on_node_uuid="4" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs>
+ <trigger>
+ <pseudo_event id="9" operation="start" operation_key="test_start_0"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="1">
+ <action_set>
+ <pseudo_event id="9" operation="start" operation_key="test_start_0">
+ <attributes CRM_meta_name="start" CRM_meta_timeout="20000" />
+ </pseudo_event>
+ </action_set>
+ <inputs>
+ <trigger>
+ <rsc_op id="8" operation="stop" operation_key="test_stop_0" on_node="rhel8-3" on_node_uuid="3"/>
+ </trigger>
+ </inputs>
+ </synapse>
+ <synapse id="2">
+ <action_set>
+ <rsc_op id="8" operation="stop" operation_key="test_stop_0" on_node="rhel8-3" on_node_uuid="3">
+ <primitive id="test" class="ocf" provider="pacemaker" type="Dummy"/>
+ <attributes CRM_meta_name="stop" CRM_meta_on_node="rhel8-3" CRM_meta_on_node_uuid="3" CRM_meta_timeout="20000" />
+ </rsc_op>
+ </action_set>
+ <inputs/>
+ </synapse>
+</transition_graph>
diff --git a/cts/scheduler/scores/stop-unexpected-2.scores b/cts/scheduler/scores/stop-unexpected-2.scores
new file mode 100644
index 000000000..0eb549f5e
--- /dev/null
+++ b/cts/scheduler/scores/stop-unexpected-2.scores
@@ -0,0 +1,21 @@
+
+pcmk__native_allocate: Fencing allocation score on rhel8-1: 0
+pcmk__native_allocate: Fencing allocation score on rhel8-2: 0
+pcmk__native_allocate: Fencing allocation score on rhel8-3: 0
+pcmk__native_allocate: Fencing allocation score on rhel8-4: 0
+pcmk__native_allocate: Fencing allocation score on rhel8-5: 0
+pcmk__native_allocate: FencingFail allocation score on rhel8-1: 0
+pcmk__native_allocate: FencingFail allocation score on rhel8-2: 0
+pcmk__native_allocate: FencingFail allocation score on rhel8-3: 0
+pcmk__native_allocate: FencingFail allocation score on rhel8-4: 0
+pcmk__native_allocate: FencingFail allocation score on rhel8-5: 0
+pcmk__native_allocate: FencingPass allocation score on rhel8-1: 0
+pcmk__native_allocate: FencingPass allocation score on rhel8-2: 0
+pcmk__native_allocate: FencingPass allocation score on rhel8-3: 0
+pcmk__native_allocate: FencingPass allocation score on rhel8-4: 0
+pcmk__native_allocate: FencingPass allocation score on rhel8-5: 0
+pcmk__native_allocate: test allocation score on rhel8-1: 0
+pcmk__native_allocate: test allocation score on rhel8-2: 0
+pcmk__native_allocate: test allocation score on rhel8-3: 0
+pcmk__native_allocate: test allocation score on rhel8-4: 0
+pcmk__native_allocate: test allocation score on rhel8-5: 0
diff --git a/cts/scheduler/summary/stop-unexpected-2.summary b/cts/scheduler/summary/stop-unexpected-2.summary
new file mode 100644
index 000000000..d6b0c15dc
--- /dev/null
+++ b/cts/scheduler/summary/stop-unexpected-2.summary
@@ -0,0 +1,29 @@
+Using the original execution date of: 2022-04-22 14:15:37Z
+Current cluster status:
+ * Node List:
+ * Online: [ rhel8-1 rhel8-2 rhel8-3 rhel8-4 rhel8-5 ]
+
+ * Full List of Resources:
+ * Fencing (stonith:fence_xvm): Started rhel8-1
+ * FencingPass (stonith:fence_dummy): Started rhel8-2
+ * FencingFail (stonith:fence_dummy): Started rhel8-3
+ * test (ocf:pacemaker:Dummy): Started [ rhel8-4 rhel8-3 ]
+
+Transition Summary:
+ * Restart test ( rhel8-4 )
+
+Executing Cluster Transition:
+ * Resource action: test stop on rhel8-3
+ * Pseudo action: test_start_0
+ * Resource action: test monitor=10000 on rhel8-4
+Using the original execution date of: 2022-04-22 14:15:37Z
+
+Revised Cluster Status:
+ * Node List:
+ * Online: [ rhel8-1 rhel8-2 rhel8-3 rhel8-4 rhel8-5 ]
+
+ * Full List of Resources:
+ * Fencing (stonith:fence_xvm): Started rhel8-1
+ * FencingPass (stonith:fence_dummy): Started rhel8-2
+ * FencingFail (stonith:fence_dummy): Started rhel8-3
+ * test (ocf:pacemaker:Dummy): Started rhel8-4
diff --git a/cts/scheduler/xml/stop-unexpected-2.xml b/cts/scheduler/xml/stop-unexpected-2.xml
new file mode 100644
index 000000000..e103629e9
--- /dev/null
+++ b/cts/scheduler/xml/stop-unexpected-2.xml
@@ -0,0 +1,204 @@
+<cib crm_feature_set="3.13.0" validate-with="pacemaker-3.7" epoch="59" num_updates="14" admin_epoch="0" cib-last-written="Fri Apr 22 09:15:36 2022" update-origin="rhel8-1" update-client="crmd" update-user="hacluster" have-quorum="1" dc-uuid="2" execution-date="1650636937">
+ <configuration>
+ <crm_config>
+ <cluster_property_set id="cib-bootstrap-options">
+ <nvpair id="cts-stonith-enabled" name="stonith-enabled" value="1"/>
+ <nvpair id="cts-start-failure-is-fatal" name="start-failure-is-fatal" value="false"/>
+ <nvpair id="cts-pe-input-series-max" name="pe-input-series-max" value="5000"/>
+ <nvpair id="cts-shutdown-escalation" name="shutdown-escalation" value="5min"/>
+ <nvpair id="cts-batch-limit" name="batch-limit" value="10"/>
+ <nvpair id="cts-dc-deadtime" name="dc-deadtime" value="5s"/>
+ <nvpair id="cts-no-quorum-policy" name="no-quorum-policy" value="stop"/>
+ <nvpair id="cib-bootstrap-options-have-watchdog" name="have-watchdog" value="false"/>
+ <nvpair id="cib-bootstrap-options-dc-version" name="dc-version" value="2.1.2-4.el8_6.1-ada5c3b36e2"/>
+ <nvpair id="cib-bootstrap-options-cluster-infrastructure" name="cluster-infrastructure" value="corosync"/>
+ <nvpair id="cib-bootstrap-options-cluster-name" name="cluster-name" value="rhel8-lab"/>
+ <nvpair id="cib-bootstrap-options-last-lrm-refresh" name="last-lrm-refresh" value="1650636936"/>
+ </cluster_property_set>
+ </crm_config>
+ <nodes>
+ <node id="1" uname="rhel8-1"/>
+ <node id="3" uname="rhel8-3"/>
+ <node id="4" uname="rhel8-4"/>
+ <node id="5" uname="rhel8-5"/>
+ <node id="2" uname="rhel8-2"/>
+ </nodes>
+ <resources>
+ <primitive class="stonith" id="Fencing" type="fence_xvm">
+ <meta_attributes id="Fencing-meta">
+ <nvpair id="Fencing-migration-threshold" name="migration-threshold" value="5"/>
+ </meta_attributes>
+ <instance_attributes id="Fencing-params">
+ <nvpair id="Fencing-pcmk_host_map" name="pcmk_host_map" value="remote-rhel8-1:rhel8-1;remote-rhel8-2:rhel8-2;remote-rhel8-3:rhel8-3;remote-rhel8-4:rhel8-4;remote-rhel8-5:rhel8-5;"/>
+ <nvpair id="Fencing-key_file" name="key_file" value="/etc/pacemaker/fence_xvm.key"/>
+ <nvpair id="Fencing-multicast_address" name="multicast_address" value="239.255.100.100"/>
+ <nvpair id="Fencing-pcmk_host_list" name="pcmk_host_list" value="rhel8-1 remote-rhel8-1 rhel8-2 remote-rhel8-2 rhel8-3 remote-rhel8-3 rhel8-4 remote-rhel8-4 rhel8-5 remote-rhel8-5"/>
+ </instance_attributes>
+ <operations>
+ <op id="Fencing-monitor-120s" interval="120s" name="monitor" timeout="120s"/>
+ <op id="Fencing-stop-0" interval="0" name="stop" timeout="60s"/>
+ <op id="Fencing-start-0" interval="0" name="start" timeout="60s"/>
+ </operations>
+ </primitive>
+ <primitive class="stonith" id="FencingPass" type="fence_dummy">
+ <instance_attributes id="FencingPass-params">
+ <nvpair id="FencingPass-pcmk_host_list" name="pcmk_host_list" value="rhel8-4 remote-rhel8-4 rhel8-5 remote-rhel8-5"/>
+ <nvpair id="FencingPass-random_sleep_range" name="random_sleep_range" value="30"/>
+ <nvpair id="FencingPass-mode" name="mode" value="pass"/>
+ </instance_attributes>
+ </primitive>
+ <primitive class="stonith" id="FencingFail" type="fence_dummy">
+ <instance_attributes id="FencingFail-params">
+ <nvpair id="FencingFail-pcmk_host_list" name="pcmk_host_list" value="rhel8-2 remote-rhel8-2"/>
+ <nvpair id="FencingFail-random_sleep_range" name="random_sleep_range" value="30"/>
+ <nvpair id="FencingFail-mode" name="mode" value="fail"/>
+ </instance_attributes>
+ </primitive>
+ <primitive class="ocf" id="test" provider="pacemaker" type="Dummy">
+ <meta_attributes id="test-meta_attributes">
+ <nvpair id="test-meta_attributes-multiple-active" name="multiple-active" value="stop_unexpected"/>
+ </meta_attributes>
+ <operations>
+ <op id="test-migrate_from-interval-0s" interval="0s" name="migrate_from" timeout="20s"/>
+ <op id="test-migrate_to-interval-0s" interval="0s" name="migrate_to" timeout="20s"/>
+ <op id="test-monitor-interval-10s" interval="10s" name="monitor" timeout="20s"/>
+ <op id="test-reload-interval-0s" interval="0s" name="reload" timeout="20s"/>
+ <op id="test-reload-agent-interval-0s" interval="0s" name="reload-agent" timeout="20s"/>
+ <op id="test-start-interval-0s" interval="0s" name="start" timeout="20s"/>
+ <op id="test-stop-interval-0s" interval="0s" name="stop" timeout="20s"/>
+ </operations>
+ </primitive>
+ </resources>
+ <constraints/>
+ <fencing-topology>
+ <fencing-level devices="FencingFail" id="cts-rhel8-2.1" index="1" target="rhel8-2"/>
+ <fencing-level devices="Fencing" id="cts-rhel8-2.2" index="2" target="rhel8-2"/>
+ <fencing-level devices="FencingFail" id="cts-remote-rhel8-2.1" index="1" target="remote-rhel8-2"/>
+ <fencing-level devices="Fencing" id="cts-remote-rhel8-2.2" index="2" target="remote-rhel8-2"/>
+ <fencing-level devices="FencingPass,Fencing" id="cts-rhel8-4.1" index="1" target="rhel8-4"/>
+ <fencing-level devices="FencingPass,Fencing" id="cts-remote-rhel8-4.1" index="1" target="remote-rhel8-4"/>
+ <fencing-level devices="FencingPass,Fencing" id="cts-rhel8-5.1" index="1" target="rhel8-5"/>
+ <fencing-level devices="FencingPass,Fencing" id="cts-remote-rhel8-5.1" index="1" target="remote-rhel8-5"/>
+ </fencing-topology>
+ <op_defaults>
+ <meta_attributes id="cts-op_defaults-meta">
+ <nvpair id="cts-op_defaults-timeout" name="timeout" value="90s"/>
+ </meta_attributes>
+ </op_defaults>
+ <alerts>
+ <alert id="alert-1" path="/var/lib/pacemaker/notify.sh">
+ <recipient id="alert-1-recipient-1" value="/run/crm/alert.log"/>
+ </alert>
+ </alerts>
+ </configuration>
+ <status>
+ <node_state id="4" uname="rhel8-4" in_ccm="true" crmd="online" crm-debug-origin="do_update_resource" join="member" expected="member">
+ <transient_attributes id="4">
+ <instance_attributes id="status-4"/>
+ </transient_attributes>
+ <lrm id="4">
+ <lrm_resources>
+ <lrm_resource id="FencingPass" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingPass_last_0" operation_key="FencingPass_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="47:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;47:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-4" call-id="13" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="bdca24cab6ded2b426c6b31df675bf0b"/>
+ </lrm_resource>
+ <lrm_resource id="Fencing" type="fence_xvm" class="stonith">
+ <lrm_rsc_op id="Fencing_last_0" operation_key="Fencing_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="46:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;46:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-4" call-id="9" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="9" queue-time="0" op-digest="bf974d77f2d4d33e434be1f89e362a52"/>
+ </lrm_resource>
+ <lrm_resource id="FencingFail" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingFail_last_0" operation_key="FencingFail_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="48:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;48:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-4" call-id="17" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="fe7e3f8acdd3228efda2766a0eea7ba5"/>
+ </lrm_resource>
+ <lrm_resource id="test" type="Dummy" class="ocf" provider="pacemaker">
+ <lrm_rsc_op id="test_last_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="5:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;5:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-4" call-id="130" rc-code="0" op-status="0" interval="0" last-rc-change="1650636936" exec-time="25" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-secure-params=" passwd " op-secure-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ <lrm_rsc_op id="test_last_failure_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="5:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;5:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-4" call-id="130" rc-code="0" op-status="0" interval="0" last-rc-change="1650636936" exec-time="25" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ <node_state id="5" uname="rhel8-5" in_ccm="true" crmd="online" crm-debug-origin="do_update_resource" join="member" expected="member">
+ <transient_attributes id="5">
+ <instance_attributes id="status-5"/>
+ </transient_attributes>
+ <lrm id="5">
+ <lrm_resources>
+ <lrm_resource id="FencingPass" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingPass_last_0" operation_key="FencingPass_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="62:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;62:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-5" call-id="13" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="bdca24cab6ded2b426c6b31df675bf0b"/>
+ </lrm_resource>
+ <lrm_resource id="Fencing" type="fence_xvm" class="stonith">
+ <lrm_rsc_op id="Fencing_last_0" operation_key="Fencing_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="61:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;61:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-5" call-id="9" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="6" queue-time="0" op-digest="bf974d77f2d4d33e434be1f89e362a52"/>
+ </lrm_resource>
+ <lrm_resource id="FencingFail" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingFail_last_0" operation_key="FencingFail_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="63:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;63:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-5" call-id="17" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="fe7e3f8acdd3228efda2766a0eea7ba5"/>
+ </lrm_resource>
+ <lrm_resource id="test" type="Dummy" class="ocf" provider="pacemaker">
+ <lrm_rsc_op id="test_last_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="6:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;6:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-5" call-id="113" rc-code="7" op-status="0" interval="0" last-rc-change="1650636936" exec-time="25" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-secure-params=" passwd " op-secure-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ <node_state id="1" uname="rhel8-1" in_ccm="true" crmd="online" crm-debug-origin="do_update_resource" join="member" expected="member">
+ <transient_attributes id="1">
+ <instance_attributes id="status-1"/>
+ </transient_attributes>
+ <lrm id="1">
+ <lrm_resources>
+ <lrm_resource id="FencingPass" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingPass_last_0" operation_key="FencingPass_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="2:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;2:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-1" call-id="13" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="bdca24cab6ded2b426c6b31df675bf0b"/>
+ </lrm_resource>
+ <lrm_resource id="Fencing" type="fence_xvm" class="stonith">
+ <lrm_rsc_op id="Fencing_last_0" operation_key="Fencing_start_0" operation="start" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="76:0:0:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;76:0:0:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-1" call-id="14" rc-code="0" op-status="0" interval="0" last-rc-change="1650636745" exec-time="36" queue-time="0" op-digest="bf974d77f2d4d33e434be1f89e362a52"/>
+ <lrm_rsc_op id="Fencing_monitor_120000" operation_key="Fencing_monitor_120000" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="77:0:0:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;77:0:0:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-1" call-id="16" rc-code="0" op-status="0" interval="120000" last-rc-change="1650636745" exec-time="36" queue-time="0" op-digest="24c9c9364f847dcb857d6fb4e1b4d3c8"/>
+ </lrm_resource>
+ <lrm_resource id="FencingFail" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingFail_last_0" operation_key="FencingFail_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="3:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;3:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-1" call-id="21" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="fe7e3f8acdd3228efda2766a0eea7ba5"/>
+ </lrm_resource>
+ <lrm_resource id="test" type="Dummy" class="ocf" provider="pacemaker">
+ <lrm_rsc_op id="test_last_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="2:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;2:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-1" call-id="121" rc-code="7" op-status="0" interval="0" last-rc-change="1650636936" exec-time="62" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-secure-params=" passwd " op-secure-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ <node_state id="2" uname="rhel8-2" in_ccm="true" crmd="online" crm-debug-origin="do_update_resource" join="member" expected="member">
+ <transient_attributes id="2">
+ <instance_attributes id="status-2"/>
+ </transient_attributes>
+ <lrm id="2">
+ <lrm_resources>
+ <lrm_resource id="FencingPass" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingPass_last_0" operation_key="FencingPass_start_0" operation="start" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="78:0:0:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;78:0:0:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-2" call-id="18" rc-code="0" op-status="0" interval="0" last-rc-change="1650636745" exec-time="21041" queue-time="0" op-digest="bdca24cab6ded2b426c6b31df675bf0b"/>
+ </lrm_resource>
+ <lrm_resource id="Fencing" type="fence_xvm" class="stonith">
+ <lrm_rsc_op id="Fencing_last_0" operation_key="Fencing_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="16:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;16:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-2" call-id="9" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="2" queue-time="0" op-digest="bf974d77f2d4d33e434be1f89e362a52"/>
+ </lrm_resource>
+ <lrm_resource id="FencingFail" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingFail_last_0" operation_key="FencingFail_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="18:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;18:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-2" call-id="17" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="fe7e3f8acdd3228efda2766a0eea7ba5"/>
+ </lrm_resource>
+ <lrm_resource id="test" type="Dummy" class="ocf" provider="pacemaker">
+ <lrm_rsc_op id="test_last_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="3:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;3:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-2" call-id="109" rc-code="7" op-status="0" interval="0" last-rc-change="1650636936" exec-time="37" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-secure-params=" passwd " op-secure-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ <node_state id="3" uname="rhel8-3" in_ccm="true" crmd="online" crm-debug-origin="do_update_resource" join="member" expected="member">
+ <transient_attributes id="3">
+ <instance_attributes id="status-3"/>
+ </transient_attributes>
+ <lrm id="3">
+ <lrm_resources>
+ <lrm_resource id="FencingPass" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingPass_last_0" operation_key="FencingPass_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="32:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;32:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-3" call-id="13" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="0" queue-time="0" op-digest="bdca24cab6ded2b426c6b31df675bf0b"/>
+ </lrm_resource>
+ <lrm_resource id="Fencing" type="fence_xvm" class="stonith">
+ <lrm_rsc_op id="Fencing_last_0" operation_key="Fencing_monitor_0" operation="monitor" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="31:0:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:7;31:0:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-3" call-id="9" rc-code="7" op-status="0" interval="0" last-rc-change="1650636745" exec-time="2" queue-time="0" op-digest="bf974d77f2d4d33e434be1f89e362a52"/>
+ </lrm_resource>
+ <lrm_resource id="FencingFail" type="fence_dummy" class="stonith">
+ <lrm_rsc_op id="FencingFail_last_0" operation_key="FencingFail_start_0" operation="start" crm-debug-origin="build_active_RAs" crm_feature_set="3.13.0" transition-key="79:0:0:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;79:0:0:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-3" call-id="26" rc-code="0" op-status="0" interval="0" last-rc-change="1650636745" exec-time="1044" queue-time="0" op-digest="fe7e3f8acdd3228efda2766a0eea7ba5"/>
+ </lrm_resource>
+ <lrm_resource id="test" type="Dummy" class="ocf" provider="pacemaker">
+ <lrm_rsc_op id="test_last_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="4:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;4:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-3" call-id="114" rc-code="0" op-status="0" interval="0" last-rc-change="1650636936" exec-time="24" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-force-restart=" state " op-restart-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8" op-secure-params=" passwd " op-secure-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ <lrm_rsc_op id="test_last_failure_0" operation_key="test_monitor_0" operation="monitor" crm-debug-origin="do_update_resource" crm_feature_set="3.13.0" transition-key="4:46:7:181f4609-25b9-4673-b760-5de7a7f55635" transition-magic="0:0;4:46:7:181f4609-25b9-4673-b760-5de7a7f55635" exit-reason="" on_node="rhel8-3" call-id="114" rc-code="0" op-status="0" interval="0" last-rc-change="1650636936" exec-time="24" queue-time="0" op-digest="f2317cad3d54cec5d7d7aa7d0bf35cf8"/>
+ </lrm_resource>
+ </lrm_resources>
+ </lrm>
+ </node_state>
+ </status>
+</cib>
--
2.27.0
From 60d8bb01ba73dfd1cb25c6764ee2b923dcfc4e8c Mon Sep 17 00:00:00 2001
From: Ken Gaillot <kgaillot@redhat.com>
Date: Fri, 22 Apr 2022 14:09:43 -0500
Subject: [PATCH 3/3] Revert "Refactor: scheduler: add expected node to
primitive variant data"
This reverts commit 871e2201d92520039df45062afc9120fd1fb0f30.
---
include/crm/pengine/internal.h | 4 ----
lib/pengine/native.c | 38 ----------------------------------
lib/pengine/variant.h | 8 ++-----
3 files changed, 2 insertions(+), 48 deletions(-)
diff --git a/include/crm/pengine/internal.h b/include/crm/pengine/internal.h
index a2e4b5bf7..fe9a23b7e 100644
--- a/include/crm/pengine/internal.h
+++ b/include/crm/pengine/internal.h
@@ -580,8 +580,4 @@ xmlNode *pe__failed_probe_for_rsc(pe_resource_t *rsc, const char *name);
const char *pe__clone_child_id(pe_resource_t *rsc);
-void pe__update_expected_node(pe_resource_t *rsc, pe_node_t *node,
- int execution_status, int exit_status,
- int expected_exit_status);
-
#endif
diff --git a/lib/pengine/native.c b/lib/pengine/native.c
index 591d1c6f5..fa7dc8960 100644
--- a/lib/pengine/native.c
+++ b/lib/pengine/native.c
@@ -1376,41 +1376,3 @@ pe__native_is_filtered(pe_resource_t *rsc, GList *only_rsc, gboolean check_paren
return TRUE;
}
-
-/*!
- * \internal
- * \brief Set a resource's expected node if appropriate for a history result
- *
- * \param[in] rsc Resource to set expected node for
- * \param[in] node Node to set as expected node
- * \param[in] execution_status History entry's execution status
- * \param[in] exit_status History entry's actual exit status
- * \param[in] expected_status History entry's expected exit status
- */
-void
-pe__update_expected_node(pe_resource_t *rsc, pe_node_t *node,
- int execution_status, int exit_status,
- int expected_exit_status)
-{
- native_variant_data_t *native_data = NULL;
-
- get_native_variant_data(native_data, rsc);
-
- if ((rsc->recovery_type == recovery_stop_unexpected)
- && (rsc->role > RSC_ROLE_STOPPED)
- && (execution_status == PCMK_EXEC_DONE)
- && (exit_status == expected_exit_status)) {
- // Resource is active and was expected on this node
- pe_rsc_trace(rsc, "Found expected node %s for %s",
- node->details->uname, rsc->id);
- native_data->expected_node = node;
- pe__set_resource_flags(rsc, pe_rsc_stop_unexpected);
-
- } else if ((native_data->expected_node != NULL)
- && (native_data->expected_node->details == node->details)) {
- // Resource is not cleanly active here
- pe_rsc_trace(rsc, "Clearing expected node for %s", rsc->id);
- native_data->expected_node = NULL;
- pe__clear_resource_flags(rsc, pe_rsc_stop_unexpected);
- }
-}
diff --git a/lib/pengine/variant.h b/lib/pengine/variant.h
index d8fefa9d6..cabfbe81f 100644
--- a/lib/pengine/variant.h
+++ b/lib/pengine/variant.h
@@ -1,5 +1,5 @@
/*
- * Copyright 2004-2022 the Pacemaker project contributors
+ * Copyright 2004-2021 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
@@ -139,11 +139,7 @@ typedef struct group_variant_data_s {
# elif VARIANT_NATIVE
typedef struct native_variant_data_s {
- /* If the resource is multiply active, and has multiple-active set to
- * stop_unexpected, this will be set to the node where the resource was
- * found active by an operation with a expected result.
- */
- pe_node_t *expected_node;
+ int dummy;
} native_variant_data_t;
# define get_native_variant_data(data, rsc) \
--
2.27.0

13
SPECS/pacemaker.spec
View File

@ -242,7 +242,7 @@
Name: pacemaker
Summary: Scalable High-Availability cluster resource manager
Version: %{pcmkversion}
Release: %{pcmk_release}%{?dist}
Release: %{pcmk_release}%{?dist}.2
%if %{defined _unitdir}
License: GPLv2+ and LGPLv2+
%else
@ -286,6 +286,9 @@ Patch20: 020-systemd-unit.patch
Patch21: 021-failure-messages.patch
Patch22: 022-memory-leak.patch
Patch23: 023-regression.patch
Patch24: 024-stop_unexpected.patch
Patch25: 025-stop_unexpected-test.patch
Patch26: 026-stop_unexpected-fix.patch
# downstream-only commits
#Patch1xx: 1xx-xxxx.patch
@ -983,6 +986,14 @@ exit 0
%license %{nagios_name}-%{nagios_hash}/COPYING
%changelog
* Fri Apr 22 2022 Ken Gaillot <kgaillot@redhat.com> - 2.1.2-4.2
- Fix issue with "stop_unexpected" value for "multiple-active" meta-attribute
- Resolves: rhbz2062848
* Fri Apr 8 2022 Ken Gaillot <kgaillot@redhat.com> - 2.1.2-4.1
- Support "stop_unexpected" value for "multiple-active" meta-attribute
- Resolves: rhbz2062848
* Wed Jan 26 2022 Ken Gaillot <kgaillot@redhat.com> - 2.1.2-4
- Fix regression in down event detection that affects remote nodes
- Resolves: rhbz2046446