rpms/fence-agents
7
0
Fork 0
Code Issues Pull Requests Releases Activity
db7f5a0638
fence-agents/SOURCES/bz2033671-fence_sbd-improve-error-handling.patch
CentOS Sources db7f5a0638 import fence-agents-4.2.1-100.el8
2022-08-03 10:12:35 +00:00

297 lines
10 KiB
Diff
Raw Blame History

From ab1de07902d9f380c10405d6ddac3aeb43838c86 Mon Sep 17 00:00:00 2001
From: Klaus Wenninger <klaus.wenninger@aon.at>
Date: Thu, 28 Jul 2022 15:33:12 +0200
Subject: [PATCH] fence_sbd: improve error handling
basically when using 3 disks be happy with 2 answers
but give it 5s at least to collect all answers
increase default power-timeout to 30s so that waiting
those 5s still allows us to get done sending the
reboot
RHBZ#2033671
---
agents/sbd/fence_sbd.py | 77 +++++++++++++--------
lib/fencing.py.py | 109 ++++++++++++++++++++++++++++++
tests/data/metadata/fence_sbd.xml | 2 +-
3 files changed, 158 insertions(+), 30 deletions(-)
diff --git a/agents/sbd/fence_sbd.py b/agents/sbd/fence_sbd.py
index 0c876b16e..2b0127d55 100644
--- a/agents/sbd/fence_sbd.py
+++ b/agents/sbd/fence_sbd.py
@@ -5,7 +5,7 @@
import os
import atexit
sys.path.append("@FENCEAGENTSLIBDIR@")
-from fencing import fail_usage, run_command, fence_action, all_opt
+from fencing import fail_usage, run_commands, fence_action, all_opt
from fencing import atexit_handler, check_input, process_input, show_docs
from fencing import run_delay
import itertools
@@ -81,7 +81,7 @@ def check_sbd_device(options, device_path):
cmd = "%s -d %s dump" % (options["--sbd-path"], device_path)
- (return_code, out, err) = run_command(options, cmd)
+ (return_code, out, err) = run_commands(options, [ cmd ])
for line in itertools.chain(out.split("\n"), err.split("\n")):
if len(line) == 0:
@@ -94,21 +94,35 @@ def check_sbd_device(options, device_path):
return DEVICE_INIT
+
def generate_sbd_command(options, command, arguments=None):
"""Generates a sbd command based on given arguments.
Return Value:
- generated sbd command (string)
+ generated list of sbd commands (strings) depending
+ on command multiple commands with a device each
+ or a single command with multiple devices
"""
- cmd = options["--sbd-path"]
+ cmds = []
+
+ if not command in ["list", "dump"]:
+ cmd = options["--sbd-path"]
- # add "-d" for each sbd device
- for device in parse_sbd_devices(options):
- cmd += " -d %s" % device
+ # add "-d" for each sbd device
+ for device in parse_sbd_devices(options):
+ cmd += " -d %s" % device
- cmd += " %s %s" % (command, arguments)
+ cmd += " %s %s" % (command, arguments)
+ cmds.append(cmd)
+
+ else:
+ for device in parse_sbd_devices(options):
+ cmd = options["--sbd-path"]
+ cmd += " -d %s" % device
+ cmd += " %s %s" % (command, arguments)
+ cmds.append(cmd)
- return cmd
+ return cmds
def send_sbd_message(conn, options, plug, message):
"""Sends a message to all sbd devices.
@@ -128,7 +142,7 @@ def send_sbd_message(conn, options, plug, message):
arguments = "%s %s" % (plug, message)
cmd = generate_sbd_command(options, "message", arguments)
- (return_code, out, err) = run_command(options, cmd)
+ (return_code, out, err) = run_commands(options, cmd)
return (return_code, out, err)
@@ -147,7 +161,7 @@ def get_msg_timeout(options):
cmd = generate_sbd_command(options, "dump")
- (return_code, out, err) = run_command(options, cmd)
+ (return_code, out, err) = run_commands(options, cmd)
for line in itertools.chain(out.split("\n"), err.split("\n")):
if len(line) == 0:
@@ -288,7 +302,7 @@ def get_node_list(conn, options):
cmd = generate_sbd_command(options, "list")
- (return_code, out, err) = run_command(options, cmd)
+ (return_code, out, err) = run_commands(options, cmd)
for line in out.split("\n"):
if len(line) == 0:
@@ -356,6 +370,7 @@ def main():
all_opt["method"]["default"] = "cycle"
all_opt["method"]["help"] = "-m, --method=[method] Method to fence (onoff|cycle) (Default: cycle)"
+ all_opt["power_timeout"]["default"] = "30"
options = check_input(device_opt, process_input(device_opt))
@@ -376,23 +391,27 @@ def main():
# We need to check if the provided sbd_devices exists. We need to do
# that for every given device.
- for device_path in parse_sbd_devices(options):
- logging.debug("check device \"%s\"", device_path)
-
- return_code = check_sbd_device(options, device_path)
- if PATH_NOT_EXISTS == return_code:
- logging.error("\"%s\" does not exist", device_path)
- elif PATH_NOT_BLOCK == return_code:
- logging.error("\"%s\" is not a valid block device", device_path)
- elif DEVICE_NOT_INIT == return_code:
- logging.error("\"%s\" is not initialized", device_path)
- elif DEVICE_INIT != return_code:
- logging.error("UNKNOWN error while checking \"%s\"", device_path)
-
- # If we get any error while checking the device we need to exit at this
- # point.
- if DEVICE_INIT != return_code:
- exit(return_code)
+ # Just for the case we are really rebooting / powering off a device
+ # (pacemaker as well uses the list command to generate a dynamic list)
+ # we leave it to sbd to try and decide if it was successful
+ if not options["--action"] in ["reboot", "off", "list"]:
+ for device_path in parse_sbd_devices(options):
+ logging.debug("check device \"%s\"", device_path)
+
+ return_code = check_sbd_device(options, device_path)
+ if PATH_NOT_EXISTS == return_code:
+ logging.error("\"%s\" does not exist", device_path)
+ elif PATH_NOT_BLOCK == return_code:
+ logging.error("\"%s\" is not a valid block device", device_path)
+ elif DEVICE_NOT_INIT == return_code:
+ logging.error("\"%s\" is not initialized", device_path)
+ elif DEVICE_INIT != return_code:
+ logging.error("UNKNOWN error while checking \"%s\"", device_path)
+
+ # If we get any error while checking the device we need to exit at this
+ # point.
+ if DEVICE_INIT != return_code:
+ exit(return_code)
# we check against the defined timeouts. If the pacemaker timeout is smaller
# then that defined within sbd we should report this.
diff --git a/lib/fencing.py.py b/lib/fencing.py.py
index b746ede8b..fc3679e33 100644
--- a/lib/fencing.py.py
+++ b/lib/fencing.py.py
@@ -1088,6 +1088,115 @@ def is_executable(path):
return True
return False
+def run_commands(options, commands, timeout=None, env=None, log_command=None):
+ # inspired by psutils.wait_procs (BSD License)
+ def check_gone(proc, timeout):
+ try:
+ returncode = proc.wait(timeout=timeout)
+ except subprocess.TimeoutExpired:
+ pass
+ else:
+ if returncode is not None or not proc.is_running():
+ proc.returncode = returncode
+ gone.add(proc)
+
+ if timeout is None and "--power-timeout" in options:
+ timeout = options["--power-timeout"]
+ if timeout == 0:
+ timeout = None
+ if timeout is not None:
+ timeout = float(timeout)
+
+ time_start = time.time()
+ procs = []
+ status = None
+ pipe_stdout = ""
+ pipe_stderr = ""
+
+ for command in commands:
+ logging.info("Executing: %s\n", log_command or command)
+
+ try:
+ process = subprocess.Popen(shlex.split(command), stdout=subprocess.PIPE, stderr=subprocess.PIPE, env=env,
+ # decodes newlines and in python3 also converts bytes to str
+ universal_newlines=(sys.version_info[0] > 2))
+ except OSError:
+ fail_usage("Unable to run %s\n" % command)
+
+ procs.append(process)
+
+ gone = set()
+ alive = set(procs)
+
+ while True:
+ if alive:
+ max_timeout = 2.0 / len(alive)
+ for proc in alive:
+ if timeout is not None:
+ if time.time()-time_start >= timeout:
+ # quickly go over the rest
+ max_timeout = 0
+ check_gone(proc, max_timeout)
+ alive = alive - gone
+
+ if not alive:
+ break
+
+ if time.time()-time_start < 5.0:
+ # give it at least 5s to get a complete answer
+ # afterwards we're OK with a quorate answer
+ continue
+
+ if len(gone) > len(alive):
+ good_cnt = 0
+ for proc in gone:
+ if proc.returncode == 0:
+ good_cnt += 1
+ # a positive result from more than half is fine
+ if good_cnt > len(procs)/2:
+ break
+
+ if timeout is not None:
+ if time.time() - time_start >= timeout:
+ logging.debug("Stop waiting after %s\n", str(timeout))
+ break
+
+ logging.debug("Done: %d gone, %d alive\n", len(gone), len(alive))
+
+ for proc in gone:
+ if (status != 0):
+ status = proc.returncode
+ # hand over the best status we have
+ # but still collect as much stdout/stderr feedback
+ # avoid communicate as we know already process
+ # is gone and it seems to block when there
+ # are D state children we don't get rid off
+ os.set_blocking(proc.stdout.fileno(), False)
+ os.set_blocking(proc.stderr.fileno(), False)
+ try:
+ pipe_stdout += proc.stdout.read()
+ except:
+ pass
+ try:
+ pipe_stderr += proc.stderr.read()
+ except:
+ pass
+ proc.stdout.close()
+ proc.stderr.close()
+
+ for proc in alive:
+ proc.kill()
+
+ if status is None:
+ fail(EC_TIMED_OUT, stop=(int(options.get("retry", 0)) < 1))
+ status = EC_TIMED_OUT
+ pipe_stdout = ""
+ pipe_stderr = "timed out"
+
+ logging.debug("%s %s %s\n", str(status), str(pipe_stdout), str(pipe_stderr))
+
+ return (status, pipe_stdout, pipe_stderr)
+
def run_command(options, command, timeout=None, env=None, log_command=None):
if timeout is None and "--power-timeout" in options:
timeout = options["--power-timeout"]
diff --git a/tests/data/metadata/fence_sbd.xml b/tests/data/metadata/fence_sbd.xml
index 516370c40..7248b864a 100644
--- a/tests/data/metadata/fence_sbd.xml
+++ b/tests/data/metadata/fence_sbd.xml
@@ -87,7 +87,7 @@
</parameter>
<parameter name="power_timeout" unique="0" required="0">
<getopt mixed="--power-timeout=[seconds]" />
- <content type="second" default="20" />
+ <content type="second" default="30" />
<shortdesc lang="en">Test X seconds for status change after ON/OFF</shortdesc>
</parameter>
<parameter name="power_wait" unique="0" required="0">
Reference in New Issue View Git Blame Copy Permalink