Prometheus timeouts when querying thanos on large history range

It seems thanos returns some timeouts when requesting data on large history. For example, here, a query on system status during the last month return a timeout. After a couple of refresh the cache is getting populated and the graph content is displayed

---

Migrated from T4470 (view on Phabricator)

added Monitoring System administration ~119 labels

For info, it seems the I/Os on ceph are quite important during a large query:

assigned to @vsellier

unassigned @vsellier

I didn't had new occurrences of the issue for a while so I guess the thanos compactor helped

closed