cloud-init/SOURCES/0005-Remove-race-condition-between-cloud-init-and-Network.patch

From 386f0a82bfdfd62e506bf4251c17263260d3250a Mon Sep 17 00:00:00 2001
From: Eduardo Otubo <otubo@redhat.com>
Date: Fri, 7 May 2021 13:36:14 +0200
Subject: Remove race condition between cloud-init and NetworkManager

Message-id: <20200302104635.11648-1-otubo@redhat.com>
Patchwork-id: 94098
O-Subject: [RHEL-7.9/RHEL-8.2.0 cloud-init PATCH] Remove race condition between cloud-init and NetworkManager
Bugzilla: 1807797
RH-Acked-by: Cathy Avery <cavery@redhat.com>
RH-Acked-by: Mohammed Gamal <mgamal@redhat.com>

BZ: 1748015
BRANCH: rhel7/master-18.5
BREW: 26924611

BZ: 1807797
BRANCH: rhel820/master-18.5
BREW: 26924957

cloud-init service is set to start before NetworkManager service starts,
but this does not avoid a race condition between them. NetworkManager
starts before cloud-init can write `dns=none' to the file:
/etc/NetworkManager/conf.d/99-cloud-init.conf. This way NetworkManager
doesn't read the configuration and erases all resolv.conf values upon
shutdown. On the next reboot neither cloud-init or NetworkManager will
write anything to resolv.conf, leaving it blank.

This patch introduces a NM reload (try-restart) at the end of cloud-init
start up so it won't erase resolv.conf upon first shutdown.

x-downstream-only: yes
resolves: rhbz#1748015, rhbz#1807797 and rhbz#1804780

Signed-off-by: Eduardo Otubo <otubo@redhat.com>
Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>

This commit is a squash and also includes the folloowing commits:

commit 316a17b7c02a87fa9b2981535be0b20d165adc46
Author: Eduardo Otubo <otubo@redhat.com>
Date:   Mon Jun 1 11:58:06 2020 +0200

    Make cloud-init.service execute after network is up

    RH-Author: Eduardo Otubo <otubo@redhat.com>
    Message-id: <20200526090804.2047-1-otubo@redhat.com>
    Patchwork-id: 96809
    O-Subject: [RHEL-8.2.1 cloud-init PATCH] Make cloud-init.service execute after network is up
    Bugzilla: 1803928
    RH-Acked-by: Vitaly Kuznetsov <vkuznets@redhat.com>
    RH-Acked-by: Miroslav Rezanina <mrezanin@redhat.com>

    cloud-init.service needs to wait until network is fully up before
    continuing executing and configuring its service.

    Signed-off-by: Eduardo Otubo <otubo@redhat.com>

    x-downstream-only: yes
    Resolves: rhbz#1831646
    Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>

commit 0422ba0e773d1a8257a3f2bf3db05f3bc7917eb7
Author: Eduardo Otubo <otubo@redhat.com>
Date:   Thu May 28 08:44:08 2020 +0200

    Remove race condition between cloud-init and NetworkManager

    RH-Author: Eduardo Otubo <otubo@redhat.com>
    Message-id: <20200327121911.17699-1-otubo@redhat.com>
    Patchwork-id: 94453
    O-Subject: [RHEL-7.9/RHEL-8.2.0 cloud-init PATCHv2] Remove race condition between cloud-init and NetworkManager
    Bugzilla: 1840648
    RH-Acked-by: Vitaly Kuznetsov <vkuznets@redhat.com>
    RH-Acked-by: Miroslav Rezanina <mrezanin@redhat.com>
    RH-Acked-by: Cathy Avery <cavery@redhat.com>

    cloud-init service is set to start before NetworkManager service starts,
    but this does not avoid a race condition between them. NetworkManager
    starts before cloud-init can write `dns=none' to the file:
    /etc/NetworkManager/conf.d/99-cloud-init.conf. This way NetworkManager
    doesn't read the configuration and erases all resolv.conf values upon
    shutdown. On the next reboot neither cloud-init or NetworkManager will
    write anything to resolv.conf, leaving it blank.

    This patch introduces a NM reload (try-reload-or-restart) at the end of cloud-init
    start up so it won't erase resolv.conf upon first shutdown.

    x-downstream-only: yes

    Signed-off-by: Eduardo Otubo otubo@redhat.com
    Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>

commit e0b48a936433faea7f56dbc29dda35acf7d375f7
Author: Eduardo Otubo <otubo@redhat.com>
Date:   Thu May 28 08:44:06 2020 +0200

    Enable ssh_deletekeys by default

    RH-Author: Eduardo Otubo <otubo@redhat.com>
    Message-id: <20200317091705.15715-1-otubo@redhat.com>
    Patchwork-id: 94365
    O-Subject: [RHEL-7.9/RHEL-8.2.0 cloud-init PATCH] Enable ssh_deletekeys by default
    Bugzilla: 1814152
    RH-Acked-by: Mohammed Gamal <mgamal@redhat.com>
    RH-Acked-by: Vitaly Kuznetsov <vkuznets@redhat.com>

    The configuration option ssh_deletekeys will trigger the generation
    of new ssh keys for every new instance deployed.

    x-downstream-only: yes
    resolves: rhbz#1814152

    Signed-off-by: Eduardo Otubo <otubo@redhat.com>
    Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
---
 rhel/cloud.cfg                  | 2 +-
 rhel/systemd/cloud-init.service | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/rhel/cloud.cfg b/rhel/cloud.cfg
index 82e8bf62..9ecba215 100644
--- a/rhel/cloud.cfg
+++ b/rhel/cloud.cfg
@@ -6,7 +6,7 @@ ssh_pwauth:   0
 
 mount_default_fields: [~, ~, 'auto', 'defaults,nofail,x-systemd.requires=cloud-init.service', '0', '2']
 resize_rootfs_tmp: /dev
-ssh_deletekeys:   0
+ssh_deletekeys:   1
 ssh_genkeytypes:  ~
 syslog_fix_perms: ~
 disable_vmware_customization: false
diff --git a/rhel/systemd/cloud-init.service b/rhel/systemd/cloud-init.service
index d0023a05..0b3d796d 100644
--- a/rhel/systemd/cloud-init.service
+++ b/rhel/systemd/cloud-init.service
@@ -5,6 +5,7 @@ Wants=sshd-keygen.service
 Wants=sshd.service
 After=cloud-init-local.service
 After=NetworkManager.service network.service
+After=NetworkManager-wait-online.service
 Before=network-online.target
 Before=sshd-keygen.service
 Before=sshd.service
-- 
2.31.1