Feature #47775
limit osd_pglog size by memory
0%
Description
We have an S3 cluster with OSDs running out of memory due to the large amount of ram needed to hold 3000 pglog entries.
osd_max_pg_log_entries = 3000
osd_min_pg_log_entries = 3000
dump_mempools:
"osd_pglog": {
"items": 445728,
"bytes": 1299377741
},
(This is on an osd with osd_memory_target = 2GB, and the osd has 112 PGs).
So we would like to be able to set something like osd_pg_log_memory_limit which would automatically decrease osd_max_pg_log_entries when the osd_pglog mempool is oversized.
Thanks!
History
#1 Updated by Dan van der Ster over 3 years ago
I've seen that octopus has a new option osd_target_pg_log_entries_per_osd. This looks good. I would therefore amend my idea above to autotune osd_target_pg_log_entries_per_osd according to the osd_pglog mempool.
It might be a good idea to relates this to the osd_memory_target: (e.g. osd_memory_pglog_target_ratio=0.1 by default).
#2 Updated by Greg Farnum almost 3 years ago
- Project changed from Ceph to RADOS