Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>

Signed-off-by: Simon Pasquier <spasquie@redhat.com>

Signed-off-by: Simon Pasquier <spasquie@redhat.com>

There are still some dependencies that we need to make work to fully
deactivate the legacyImports in the future. I'll start opening PRs
against those other repositories.

Signed-off-by: Damien Grisonnet <dgrisonn@redhat.com>

This daemonset doesn't affect workload availability so allow its rollout to
be parallelized.

This could have been achieved either by switching to stringData, or doing
`std.base64(std.encodeUTF8($._config.alertmanager.config))` as per
google/jsonnet#575

I went with the former, because it's:
1. Easier to read existing config
2. Consistent with the way jsonnet object-based config is written just above

Expose the Thanos Sidecar's HTTP metrics and gRPC StoreAPI interfaces via a dedicated service. ClusterIP is set to none to allow for full discovery of all endpoints behind the service via in-cluster DNS. This allows for a new ServiceMonitor to then scrape the metrics available on the Sidecar's HTTP port.

A new service has been used so as to separate out the metrics that Prometheus makes available, and the metrics the Thanos Sidecar makes available. The Thanos Mixins from thanos-io/thanos default to a job label of 'thanos-sidecar', and hence the service here has had this label applied.

Watch error alert

kube-apiserver has a histogram etcd_request_duration_seconds that
measures latency between the kube-apiserver and etcd instance.
This metrics is currently dropped by cluster-prometheus. Enable
this metrics so we have visibility into etcd latency.

We ensured that this does not enable other unwanted metrcis

count by(name) ({name=~"etcd_request.+"})

etcd_request_duration_seconds_bucket
etcd_request_duration_seconds_count
etcd_request_duration_seconds_sum