No restart JIPP capability

In the old days the Portal offered a restart capability so that flaky JIPPs or their predecessors could be restarted.

This facility seems to have gone and is particularly necessary now that JIPPs are so flaky.

changed the description

added prioritymedium service:jenkins severitynormal statetodo team:releng labels

marked this issue as related to #4379 (closed)

added Enhancement label

+1

I'd like to see this functionality return, in some form.

-1

In the old days, this was never a reliable solution and only a workaround for underlying problems. We deliberately chose to not port this functionality when we migrated our CI instances to our cluster environment. For the most part, restarting a JIPP became a LOT LESS of an issue.

Re-adding this functionality would be again a workaround for underlying problems. In many cases, a simple restart does not solve all issues and requires manual intervention anyway. So I'd rather invest time and effort to get rid of the real problems.

ok. No point in a facility that doesn't work.

closed

removed statetodo label

"restart" shows up in 51 HelpDesk issues.

The search term "restart" is not precise enough.

Out of those 51 HelpDesk "restart" issues:

• 31 are not requests to restart Jenkins
• 2 are about adding the restart functionality
• 8 requests to restart Jenkins, did not help, required manual intervention
• 1 request to restart Jenkins, but restart was not required
• 9 requests to restart Jenkins, where a simple restart actually helped

Apart from that, the search functionality in GitLab issues is completely useless and can not be trusted.

For comparison, Thunderbird finds the following in my HelpDesk email folder:

• 298 messages for "restart" (GitLab search 51)
• 114 messages for "restart Jenkins" (GitLab search 21)
• 48 messages for "restart JIPP" (GitLab search 4)

marked this issue as related to #467 (closed)

#4039 (closed) restarted 2 days ago, 3 restarts in 3 months

#4338 (closed) 1 restart yesterday

#4381 (closed) restarted 3 days ago

#4400 (closed) restarted 5 hours ago

Re-adding this functionality would be again a workaround for underlying problems. In many cases, a simple restart does not solve all issues and requires manual intervention anyway. So I'd rather invest time and effort to get rid of the real problems.

The underlying problems have not been identified for many years, and realistically will continue that trend. This leaves committers stuck and powerless as they wait for EF intervention.

For the most part, restarting a JIPP became a LOT LESS of an issue.

I'd like to reopen this issue so I can gather some real data, not just anectodal information.

reopened

The underlying problems have not been identified for many years.

We are aware of two known issues that probably cause(d) 95% of all "restart requests":

• #682 (Race condition in the Jenkins Kubernetes Plugin, which leads to agents no longer spawning)
• OKD nodes failing (for whatever reason) and stuck JIPP pods that are not automatically rescheduled to a different node.

This leaves committers stuck and powerless as they wait for EF intervention.

We are trying hard to react promptly (even outside of working hours).

Very much appreciated, and it would be great if you didn't feel like you had to do that outside of work hours.

Restart #4489 (comment 2036119) Restart #2660 (comment 2036357) Restart #4496 (comment 2036347)

mentioned in issue #4083

Just to document: this is still a thing that happens on a regular basis.

#5183 (comment 2839454)

In this case, restart functionality would not have helped. It required manual intervention due to a known bug in the Jenkins Kubernetes plugin.

mentioned in issue #5227