Close all existing processes when restarting Manticore #2798
Comments
It already does. A new instance shouldn’t start if an old one is still running. What you’re describing sounds like a bug. Could you share your searchd log and provide more details about when this issue occurred? This will help us investigate further. |
sanikolaev
added
waiting
|
It does seem the old instance of Manticore crashed and left unresponsive processes during a precaching step, which would indeed be a bug. Also I've created very large tables, so it might have been resource exhaustion of some kind. However, I wanted to bring attention more to the fact that it should not even be possible to have old and new instances of Manticore coexist, even if there had been a crash in the past. In other words, Manticore should either forcibly clean up old processes even if they are misbehaving, or refuse to start up until they are closed. When looking at In The relevant The relevant My |
That's from November 11 to 13. Do you have the log from November 28? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Proposal:
I've encountered strange behavior multiple times which ultimately was caused by old instances of Manticore search still staying alive even though I've already used
sudo systemctl restart manticore. Usually the existence of both new and old Manticore processes manifests as MYSQL and HTTP clients failing to connect to the backend.In the end, I had to first run
sudo systemctl stop manticoreto kill the new processes, then runtop -u manticoreto find the PID of the old processes, then usingsudo kill -9 [pid]on the old processes (sometimes they don't respond unless SIGKILL is used).I believe Manticore should check for the old processes itself and close them without needing manual user action. I think there may even be a risk of data loss if two instances of Manticore are simultaneously active.
Checklist:
To be completed by the assignee. Check off tasks that have been completed or are not applicable.
The text was updated successfully, but these errors were encountered: