You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
put the server process in the background. We are using qlever in a kubernetes cluster and would like to just run qlever index && qlever start. Given that kubernetes marks the container as completed when the server process is moved to the background we need to run a loop to keep the container up.
Could we maybe add another system option called native_foreground (or similar) and change 101 ff in start.py to something along the lines of:
Didnt test this but I suspect the start_cmd without nohup would keep the server process in the foreground and prevent kubernetes from marking it as completed.
The text was updated successfully, but these errors were encountered:
You could use qlever index && qlever start && qlever log as a workaround. This has the added benefit that you get the logs on stdout. If you don't need that you could also use the usual tricks (like tail -f /dev/null, sleep infinity or cat) to keep a container running.
yes, thats what we are doing for now. However, I think it would be cleaner to have the server process just run in the foreground of the container. I'll give qlever log a try. Given thats it not a super important issue and its not preventing us from using the service you can close the issue. However, I would suggest to either remove the nohup from the native start command (its easier to just wrap qlever start in nohup than the other way round) or add a foreground option as suggested above.
Since ad-freiburg/qlever#1439 has been merged its possible to run qlever-control directly within the container. Thats great, but currently these lines
qlever-control/src/qlever/commands/start.py
Lines 112 to 113 in 547add9
qlever index && qlever start
. Given that kubernetes marks the container as completed when the server process is moved to the background we need to run a loop to keep the container up.Could we maybe add another system option called
native_foreground
(or similar) and change 101 ff in start.py to something along the lines of:Didnt test this but I suspect the
start_cmd
without nohup would keep the server process in the foreground and prevent kubernetes from marking it as completed.The text was updated successfully, but these errors were encountered: