-
Notifications
You must be signed in to change notification settings - Fork 652
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[config] Eliminate race condition between reloading Monit config and monitoring container_checker #1543
Conversation
enabling container_checker. Signed-off-by: Yong Zhao <[email protected]>
@yozhao101: Please add the "Request for " label if requesting a cherry-pick. the "Included in label is used to indicate cherry-pick has been done, or if a PR is targeted at a release branch. |
/azp run |
Azure Pipelines successfully started running 1 pipeline(s). |
Thank Joe! I will pay attention to such issue in future. |
…1543) Signed-off-by: Yong Zhao [email protected] What I did Nightly test found a failure when we ran the command sudo config reload/load_minigraph, The error message is: admin@str-a7050-acs-1:~$ sudo config reload -y Disabling container monitoring ... Stopping SONiC target ... Running command: /usr/local/bin/sonic-cfggen -j /etc/sonic/init_cfg.json -j /etc/sonic/config_db.json --write-to-db Running command: /usr/local/bin/db_migrator.py -o migrate Resetting failed status on bgp.service Resetting failed status on caclmgrd.service Resetting failed status on dhcp_relay.service Resetting failed status on hostcfgd.service Resetting failed status on hostname-config.service Resetting failed status on interfaces-config.service Resetting failed status on lldp.service Resetting failed status on ntp-config.service Resetting failed status on pmon.service Resetting failed status on procdockerstatsd.service Resetting failed status on radv.service Resetting failed status on rsyslog-config.service Resetting failed status on swss.service Resetting failed status on syncd.service Resetting failed status on teamd.service Resetting failed status on telemetry.timer Restarting SONiC target ... Reloading Monit configuration ... Reinitializing monit daemon Enabling container monitoring ... Unix socket /var/run/monit.sock connection error -- No such file or directory The root reason is that there exists an implicit race condition between the command sudo monit reload at line 701 and the command sudo monit monitor container_checker at line 706. Both commands need access the monit.sock socket file under the directory /var/run/. Specifically the sudo monit reload at line 701 will re-initialize the Monit daemon, delete old monit.sock file and then create a new one. During this re-initializing process, the command sudo monit status can always execute successfully at line 704 before the old monit.sock file was deleted, but the command sudo monit monitor container_checker at line 706 will only succeed if the new monit.sock was created, otherwise it will fail and raise this error message. How I did it I changed the sequence between the operation to reload Monit configuration and the operation to enable monitoring container_checker. How to verify it I verified this change on DuT str-a7050-acs-1 by running the command sudo config reload/load_minigraph -y to make sure the error was not raised again. Previous command output (if the output of a command-line utility has changed) admin@str-a7050-acs-1:~$ sudo config reload -y Disabling container monitoring ... Stopping SONiC target ... Running command: /usr/local/bin/sonic-cfggen -j /etc/sonic/init_cfg.json -j /etc/sonic/config_db.json --write-to-db Running command: /usr/local/bin/db_migrator.py -o migrate Resetting failed status on bgp.service Resetting failed status on caclmgrd.service Resetting failed status on dhcp_relay.service Resetting failed status on hostcfgd.service Resetting failed status on hostname-config.service Resetting failed status on interfaces-config.service Resetting failed status on lldp.service Resetting failed status on ntp-config.service Resetting failed status on pmon.service Resetting failed status on procdockerstatsd.service Resetting failed status on radv.service Resetting failed status on rsyslog-config.service Resetting failed status on swss.service Resetting failed status on syncd.service Resetting failed status on teamd.service Resetting failed status on telemetry.timer Restarting SONiC target ... Reloading Monit configuration ... Reinitializing monit daemon Enabling container monitoring ... New command output (if the output of a command-line utility has changed) admin@str-a7050-acs-1:~$ sudo config reload -y Disabling container monitoring ... Stopping SONiC target ... Running command: /usr/local/bin/sonic-cfggen -j /etc/sonic/init_cfg.json -j /etc/sonic/config_db.json --write-to-db Running command: /usr/local/bin/db_migrator.py -o migrate Resetting failed status on bgp.service Resetting failed status on caclmgrd.service Resetting failed status on dhcp_relay.service Resetting failed status on hostcfgd.service Resetting failed status on hostname-config.service Resetting failed status on interfaces-config.service Resetting failed status on lldp.service Resetting failed status on ntp-config.service Resetting failed status on pmon.service Resetting failed status on procdockerstatsd.service Resetting failed status on radv.service Resetting failed status on rsyslog-config.service Resetting failed status on swss.service Resetting failed status on syncd.service Resetting failed status on teamd.service Resetting failed status on telemetry.timer Restarting SONiC target ... Enabling container monitoring ... Reloading Monit configuration ... Reinitializing monit daemon
Signed-off-by: Yong Zhao [email protected]
What I did
Nightly test found a failure when we ran the command
sudo config reload/load_minigraph
, The error message is:The root reason is that there exists an
implicit race condition
between the commandsudo monit reload
at line 701 andthe command
sudo monit monitor container_checker
at line 706. Both commands need access themonit.sock
socket fileunder the directory
/var/run/
.Specifically the
sudo monit reload
at line 701 will re-initialize the Monit daemon, delete oldmonit.sock
file and then create a new one. During this re-initializing process, the commandsudo monit status
can always execute successfully at line 704 before the oldmonit.sock
file was deleted, but the commandsudo monit monitor container_checker
at line 706 will only succeed if the newmonit.sock
was created, otherwise it will fail and raise this error message.How I did it
I changed the sequence between the operation to reload Monit configuration and the operation to enable monitoring container_checker.
How to verify it
I verified this change on DuT
str-a7050-acs-1
by running the commandsudo config reload/load_minigraph -y
to make sure the error was not raised again.Previous command output (if the output of a command-line utility has changed)
New command output (if the output of a command-line utility has changed)