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
If you are dynamically setting labels, never use a label which can have unbounded or infinite values. This will always result in big problems for Loki.
I believe (according to the same doc) that the correct solution would be to not label them and use a regex filtering to get logs for the specific run.
The text was updated successfully, but these errors were encountered:
@mmurto As part of fixing #168 this will have to be implemented in a more generic way. AFAIK the run_id labels are currently only added in our internal configuration based on the labels of the Kubernetes pods. With #875 it will be possible to take them directly from the log statements.
AFAIK the run_id labels are currently only added in our internal configuration based on the labels of the Kubernetes pods.
We do that as well, but the problem is more with the Logaccess. The code I linked above filters the correct logs for the run with {run_id="$ortRunId"}, which as far as I understand only filters them based on the labels, so if the label wasn't created in deployment setup, logaccess wouldn't return any logs.
The current logaccess from Loki requires the
runId
to be a label:ort-server/logaccess/loki/src/main/kotlin/LokiLogFileProvider.kt
Line 127 in 3f1f5bc
This does not seem to be advised with Loki, according to Loki's docs:
I believe (according to the same doc) that the correct solution would be to not label them and use a regex filtering to get logs for the specific run.
The text was updated successfully, but these errors were encountered: