race condition in haproxy reload #5657
Conversation
|
@a13m review please |
|
@bparees I think this is almost right. I would remove the lock before wait_to_start. Doing it after introduces another race condition, where the configuration may be changed after haproxy is executed, and the second reload won't happen. Was there a good reason to have wait_to_start in the locked section? |
|
looks good to me. thanks. |
|
[test][extended:cartridge] |
|
Online Test Results: SUCCESS (https://ci.dev.openshift.redhat.com/jenkins/job/test_pull_requests/8024/) |
|
Origin Test Results: FAILURE (https://ci.openshift.redhat.com/jenkins/job/test_pull_requests/2693/) |
|
Evaluated for origin up to e1f1042 |
|
[merge] |
|
Online Merge Results: SUCCESS (https://ci.dev.openshift.redhat.com/jenkins/job/merge_pull_requests/5735/) (Image: devenv_5013) |
|
Evaluated for online up to e1f1042 |
Merged by openshift-bot
No description provided.