-
I am running v0.12.4 of buildkit as a daemon in my kubernetes cluster. I am observing some issues with the garbage collection which I am not sure if it is a configuration problem or if I should raise a "real" issue. After running for some time, buildkitd will stop working properly, because the disk space ran out. I have configured the garbage collection as follows:
The garbage collection seems to work just fine in the beginning. But after some time when I check the disk usage ( So at some point,
Obviously the garbage collection can not clean up anything if there is no reclaimable record. But I was wondering, why they would be stuck in an "unreclaimable" state. Running
Does anyone have a hint what could trigger this behavior or what I could do to debug this? Related issue: #4218
|
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
After some debugging we found the root cause of the issue. Due to a bug with Gitlab's docker registry we disabled
Note the Removing this line resolves the issue for us. I am just wondering if this is a bug in |
Beta Was this translation helpful? Give feedback.
After some debugging we found the root cause of the issue. Due to a bug with Gitlab's docker registry we disabled
sbom
andprovenance
for the attestation:Note the
attest=
line. This would causebuildkit
to never mark the built layers asRecoverable
.Removing this line resolves the issue for us. I am just wondering if this is a bug in
buildkit
or if it expected behavior.