[newrelic-logging] BREAKING Bug in env values order #1408
Labels
bug
Categorizes issue or PR as related to a bug.
triage/pending
Issue or PR is pending for triage and prioritization.
Upon an attempted upgrade for the
nri-bundle
chart from5.0.67
to5.0.82
, we've started seeing one pod in the newrelic-logging deamonset running fine but others throwingcannot execute insert file ...
and other db related errors.For a bit of context, running on k8s 1.29 in EKS with :
On further inspection, it was noted that there was only one db file being created on the volume, named
'$(NODE_NAME)-fb.db'
, which is the obvious reason why one pod would work fine, as other's wouldn't be able to get the lock on the same file.We believe the issue was introduced with
newrelic-logging 1.21.0
, where this change was made.The reason why this is not working any more is because
NODE_NAME
is now defined afterFB_DB
. Here's the snippet from k8s docs that explains why this is an issue:At the moment, this is stopping us from applying updates Newrelic has notified us about resolving some vulnerabilities.
The text was updated successfully, but these errors were encountered: