Log noise "zip archive format invalid, code: 10" is seen for all non-ZIP files
In https://log.gprd.gitlab.net/goto/f3ba1e890ffd6949dc9a2f3094506602, we see we generate gitlab-zip-metadata error: zip archive format invalid, code: 10
errors on the order of 100 times a minute:
This is also reflected in the Prometheus metrics: https://prometheus-app.gprd.gitlab.net/graph?g0.range_input=1h&g0.expr=sum(rate(gitlab_workhorse_zip_subcommand_errors_total%5B5m%5D))%20by%20(error)&g0.tab=0
I think the problem here is that we attempt to run gitlab-zip-metadata
on all artifacts, including JSON, .gz, etc. @grzesiek Does it make sense to restrict this metadata extraction to artifacts.zip
or files that have a .zip
extension? We could also squelch the error if the extension isn't ZIP.