[20:19:50] hi o11y, are there any rules of thumb for how to pick a number of Opensearch shards for an index? [20:35:20] talking about primary shards generally I'd say starting smaller and worth keeping in mind not to exceed the number of nodes in a given storage class [20:36:35] let's say you started small at 1 shard, 2 replicas, and then dumped a probably way-too-large amount of tracing data into it :) [20:38:35] if its just a one time thing it'd probably be fine to just let that age off with the daily index [20:39:11] it's not [20:45:33] hah ok, yeah we could then try something like 2 primary shards on the next daily index, or are you wanting to go back and address whats already indexed? [20:46:03] no I'm just worried about future [20:46:09] and if there were any rules-of-thumb [20:46:37] the old stuff can age out, that's fine [20:47:08] is "10–50 GB per shard" still about right? [20:48:18] oh no its much smaller now more like 500mb [20:50:49] it was up there a little while ago though, our thanos version didn't sample natively but adding the sampling to the otel exporter helped a ton [22:25:15] cdanis: some good resources I read a while back: https://docs.aws.amazon.com/opensearch-service/latest/developerguide/bp-sharding.html