Note: This bug is displayed in read-only format because
the product is no longer active in Red Hat Bugzilla.
RHEL Engineering is moving the tracking of its product development work on RHEL 6 through RHEL 9 to Red Hat Jira (issues.redhat.com). If you're a Red Hat customer, please continue to file support cases via the Red Hat customer portal. If you're not, please head to the "RHEL project" in Red Hat Jira and file new tickets here. Individual Bugzilla bugs in the statuses "NEW", "ASSIGNED", and "POST" are being migrated throughout September 2023. Bugs of Red Hat partners with an assigned Engineering Partner Manager (EPM) are migrated in late September as per pre-agreed dates. Bugs against components "kernel", "kernel-rt", and "kpatch" are only migrated if still in "NEW" or "ASSIGNED". If you cannot log in to RH Jira, please consult article #7032570. That failing, please send an e-mail to the RH Jira admins at rh-issues@redhat.com to troubleshoot your issue as a user management inquiry. The email creates a ServiceNow ticket with Red Hat. Individual Bugzilla bugs that are migrated will be moved to status "CLOSED", resolution "MIGRATED", and set with "MigratedToJIRA" in "Keywords". The link to the successor Jira issue will be found under "Links", have a little "two-footprint" icon next to it, and direct you to the "RHEL project" in Red Hat Jira (issue links are of type "https://issues.redhat.com/browse/RHEL-XXXX", where "X" is a digit). This same link will be available in a blue banner at the top of the page informing you that that bug has been migrated.
https://www.haproxy.com/blog/truly-seamless-reloads-with-haproxy-no-more-hacks/
The OpenShift layered product uses haproxy to get traffic into our cluster. We do updates to config constantly and containers are created, destroyed, or moved between nodes. Doing reloads means we end up with (literally) hundreds of 'half dead' haproxy daemons to service connections which were being held open when we reloaded the config.
We would really really like this feature. It likely means a rebase to 1.7 and a backport of the feature.
If engineering time is a significant factor in this request I may be able to get someone on my team to help.
Can we get approval for a rebase (if engineering thinks that appropriate) for 7.5?
I misspoke. We'll still have 100's of haproxy daemons, which we can handle, but we can remove our gross hack to drop syn's during reload. We are going to try to implement the 'drop empty ack' packets workaround instead of our current drop syns.
Created attachment 1326597[details]
This is a patchset that was backported onto 1.7.8-1-fc26
This is a patchset that was backported onto 1.7.8-1-fc26. I have pushed this change set into Ubuntu artful, and I figured you guys would appreciate this patches as well. We are actively running with this patchset, and it appears to have alleviated the issues referred to in this bug.
@Eric Paris
The seamless reload patchset you referred to which I backported will definitely not fix your duplicate dead/should be dying but aren't haproxy threads. This patchset does not handoff already being serviced connections between the old and new pid. AFAIU It only hands off connections that have not yet started being serviced.
I am still experiencing the other problem you mentioned where there are a ton of half-dying haproxy's getting stuck in an unknown state. You should probably open a separate bug for that, as we are trying to solve that as well.
Could someone from Redhat take a look at my patchset please?
(In reply to David Chiluk from comment #15)
> Created attachment 1326597[details]
> This is a patchset that was backported onto 1.7.8-1-fc26
>
> This is a patchset that was backported onto 1.7.8-1-fc26. I have pushed
> this change set into Ubuntu artful, and I figured you guys would appreciate
> this patches as well. We are actively running with this patchset, and it
> appears to have alleviated the issues referred to in this bug.
This does not apply to RHEL since we're using version haproxy-1.5.18. I can add the patch to recent Fedora releases that contain 1.7, if you like. Also, did you submit this upstream to see if the backport could be added to the next stable release of 1.7? That would be great. Thanks!