This message was deleted.
# opals
Slackbot
02/24/2023, 7:12 PMThis message was deleted.
g
Gage Miller
02/24/2023, 7:14 PMPossible solutions are:
1. Drop logs for large data endpoints using opa's built in log management
2. Send logs to an external service and cap the size of the logs that can be sent
o
Or Weis
02/24/2023, 7:16 PMThanks for sharing this @Gage Miller.
I think this is worth adding to our FAQ .
And we can also look at improving the log streaming from OPA.
Btw have you tried running this scenario with an external OPA (not run by OPAL-client, but still managed by it) ?
g
Gage Miller
02/24/2023, 7:17 PMI have not tried with an external opa, it may be the case that opa is crashing instead of the opal client. Something to try out for sure
Gage Miller
02/24/2023, 7:19 PM@Or Weis if it is the opal client, are there any safeties that could be put in place so it doesn't lead to full service failure? Maybe just an error log or warning saying that the opa output can't be processed.
o
Or Weis
02/24/2023, 7:22 PMWe can definitely look into that, the opal-client can potentially monitor that state of the child process more closely.
A simple thing to do regardless, is to check the healthcheck on the OPA instance and restart the client on failure there
3 Views