Ruby sampler metrics missing for sidekiq
Via @bjk-gitlab
It appears that we're having problems again with exporting sidekiq metrics to Prometheus. This issue here refers to the absence of any ruby_*
metrics as exported by RubySampler
for what it appears to be weeks. So far it looks that at least sidekiq-cluster deployments are affected (which is what gitlab.com uses), perhaps even single-process sidekiqs (to be investigated.)
We need to determine why we're not exporting these metrics properly and address the problem.
Additionally, we should think of how we can prevent this from being noticed so late in the future, e.g. by adding alerting rules that ensure these metrics are present at all times.