Bug #22288: Nodes could not update their policies anymore - Rudder - Issue Tracker

Actions

Copy link

Bug #22288

closed

Nodes could not update their policies anymore

Added by Nicolas CHARLES almost 2 years ago. Updated over 1 year ago.

Status:

Rejected

Priority:

N/A

Assignee:

Alexis Mousset

Category:

Server components

Target version:

7.2.6

Pull Request:

Severity:

Major - prevents use of part of Rudder | no simple workaround

UX impact:

I dislike using that feature

User visibility:

Operational - other Techniques | Rudder settings | Plugins

Effort required:

Priority:

Name check:

To do

Fix check:

To do

Regression:

Description

I did a session of debug with rudder server debug <ip>, and stopped it
Later on, nodes could not update their policies

logs on the server side told

Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Deactivated successfully.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Scheduled restart job, restart counter is at 6.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: Stopped CFEngine file server.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Start request repeated too quickly.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Failed with result 'start-limit-hit'.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: Failed to start CFEngine file server.
root@15-237-108-211:/var/log/rudder/webapp# service rudder-cf-serverd status
× rudder-cf-serverd.service - CFEngine file server
     Loaded: loaded (/lib/systemd/system/rudder-cf-serverd.service; enabled; vendor preset: enabled)
     Active: failed (Result: start-limit-hit) since Fri 2023-01-20 10:57:33 UTC; 1h 50min ago
    Process: 708495 ExecStart=/opt/rudder/bin/cf-serverd --graceful-detach=600 --no-fork $VERBOSITY_OPTION (code=exited, status=0/SUCCESS)
   Main PID: 708495 (code=exited, status=0/SUCCESS)
        CPU: 318ms

Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Deactivated successfully.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Scheduled restart job, restart counter is at 6.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: Stopped CFEngine file server.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Start request repeated too quickly.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: rudder-cf-serverd.service: Failed with result 'start-limit-hit'.
Jan 20 10:57:33 15-237-108-211.training.rudder.io systemd[1]: Failed to start CFEngine file server.

it was on ubuntu22, rudder 7.2.3
restarting manually rudder-cf-serverd solved the issue

Actions

Copy link

Updated by Vincent MEMBRÉ almost 2 years ago

Target version changed from 7.2.4 to 7.2.5

Actions

Copy link

Updated by Benoît PECCATTE almost 2 years ago

Severity set to Minor - inconvenience | misleading | easy workaround
UX impact set to I dislike using that feature
User visibility set to Operational - other Techniques | Rudder settings | Plugins
Priority changed from 0 to 56

Actions

Copy link

Updated by Alexis Mousset almost 2 years ago

Assignee set to Alexis Mousset
Severity changed from Minor - inconvenience | misleading | easy workaround to Major - prevents use of part of Rudder | no simple workaround
Priority changed from 56 to 75

Actions

Copy link

Updated by Alexis Mousset almost 2 years ago

What the provided logs say:

A restart was triggered (either automatically or from outside)
restart counter is at 6 means that there were 6 restart in a StartLimitIntervalSec interval.
Apparently this reaches StartLimitBurst, and the service is not started (apparently the default is 5 restarts in a 10sec period)

Now the question that remains is why we have this many restarts:

Could be because cf-serverd is unable to start (e.g. port already used), and due to the Restart=always config it will restart in loop until reaching the limit. It seems to be the most likely option. We would need to see logs above the provided lines.
Could be because there are too many restarts triggered by the graceful restart. More logs would also help see if it's what happens.

Actions

Copy link