Project

General

Profile

Actions

Bug #47929

open

Huge RAM Usage on OSD recovery

Added by Luis Felipe Domínguez Vega over 3 years ago. Updated over 3 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
2 - major
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(RADOS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

Hi, today mi Infra provider has a blackout, then the Ceph was try to
recover but are in an inconsistent state because many OSD can recover
itself because the kernel kill it by OOM. Even now one OSD that was OK,
go down by OOM killed.

Even in a server with 32GB RAM the OSD use ALL that and never recover, i
think that can be a memory leak, ceph version octopus 15.2.3

In: https://pastebin.pl/view/59089adc
You can see that buffer_anon get 32GB, but why?? all my cluster is down
because that.

this https://pastebin.pl/view/59089adc is almost the OSD going to be killed by OOM

pglog trimmed OK, but has the same behavior, log:
https://pastebin.ubuntu.com/p/dwbXtX7wTP/

Actions #1

Updated by Neha Ojha over 3 years ago

  • Project changed from Ceph to RADOS

Can you please provide the output of "ceph -s" and "ceph pg dump"?

Actions #3

Updated by Luis Felipe Domínguez Vega over 3 years ago

Well try with:

ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-<osd id> --pgid "<stuck_pg get from log>" --force --op remove

and now the OSD is running, so there are something wrong with that's PG

Actions #4

Updated by Luis Felipe Domínguez Vega over 3 years ago

Changed and used the

--op export-remove
and then
--op import
of ceph-objectstore-tool for the failing PG and now the OSD is running great.

Actions #5

Updated by Luis Felipe Domínguez Vega over 3 years ago

there are some extrange behavior because now in another failing OSD not work at all and i execute the export-remove and import again. The file is about 17 Gb of data

Actions #6

Updated by Luis Felipe Domínguez Vega over 3 years ago

Nop, not work the export-import behavior, because on recover, when need to recover that PG then OOM killed

Actions #7

Updated by Neha Ojha over 3 years ago

Can you export and upload a copy the problematic PG via ceph-post-file?

Actions #8

Updated by Luis Felipe Domínguez Vega over 3 years ago

Neha Ojha wrote:

Can you export and upload a copy the problematic PG via ceph-post-file?

there are differents PGs on differents OSD... i will try with one, but the PGs size is about 17Gb

Actions #9

Updated by Luis Felipe Domínguez Vega over 3 years ago

Neha Ojha wrote:

Can you export and upload a copy the problematic PG via ceph-post-file?

ceph-post-file: 7639cc22-79eb-426d-ac79-e704bc34ef0f

Actions

Also available in: Atom PDF