Garron Kramer

@bom-d-van Any chance you should shed some light on this? :)

@bom-d-van Its not exhausting IOps, CPU or RAM. There are 30+ TB of whisper files, so a lot! Any suggestions for improving performance?

@Civil Thank you for the response. So the most efficient approach is to add more RAM [in order to enable indexing] to the system? (I find myself wondering about the...

@Civil @bom-d-van The problem is back, and indexing is turned on - both! Any ideas? Should I look to upgrade the disks to host-based SSD? It wouldnt appear as if...

@bom-d-van go-carbon.conf ``` [common] user = "graphite" graph-prefix = "carbon.agents.{host}" metric-endpoint = "local" metric-interval = "1m0s" max-cpu = 8 [whisper] data-dir = "/data/graphite/storage/whisper" schemas-file = "/data/graphite/conf/storage-schemas.conf" aggregation-file = "/data/graphite/conf/storage-aggregation.conf" workers...

@bom-d-van I've run queries against the box whilst running iostats, and iowait barely ticks above 7%. carbonapi also seems to flatline. I see intermittent success in graph creation, but dont...

OK, so apologies for merging information from go-graphite and carbonapi, but I'm going to put everything here: Go-Graphite: ``` [2022-08-04T14:22:32.260Z] ERROR [tcp] read error {"error": "read tcp 127.0.0.1:2503->127.0.0.1:56536: i/o timeout"}...

So I cranked up the timeline to 30 days, and that totally annihilated the machine - Iowait shoots up to 70%, RES/VIRT mem shoots past 120GB+, and the kernel then...

@Civil @bom-d-van It does seem to me that go-graphite could benefit from optimisations re RAM and indexing... * does the daemon load only those metrics relevant to the query at...

> Have you tried my recommendation config here: [#479 (comment)](https://github.com/go-graphite/go-carbon/issues/479#issuecomment-1201169945) Yes, this was enabled immediately after your suggestion. It doesnt seem to have made a difference. > It's not really...