Log file Viewer

[00:56:52] <Hiro_x> hello

[17:06:56] <tigtr> I've got a system that's been humming along for a year with good/acceptable performance, but since a few days (and even maybe a week ago?) the write performance to it has dropped to a crawl

[17:07:46] <tigtr> there's 20 billion records, 8 shards, ~35 indexes (some compound) and an insert of 1000 records takes ~30 seconds -- I'm trying to insert millions of records so this is not working well at the moment -- any suggestions?

[17:10:36] <macwinner> does vmstat show high io wait times?

[17:10:45] <macwinner> also, is your disk swapping a lot?

[17:23:14] <tigtr> it's a heavy-write database, so I've (recently) configured the shards to run on 64gb of memory only -- there's plenty of memory free and I don't see any swapping either -- I'm trying to generally follow this advice: https://www.percona.com/blog/2020/05/05/tuning-mongodb-for-bulk-loads/