Project

General

Profile

Actions
Copy link

Bug #43187

open

Should increase the default CEPH_VOLUME_SYSTEMD_TRIES to 2000 to match CEPH_VOLUME_TIMEOUT(10000 seconds)

Added by dongdong tao over 4 years ago. Updated over 4 years ago.

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

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

Currently, we set CEPH_VOLUME_TIMEOUT to 10000 seoncds like below
---
Environment=CEPH_VOLUME_TIMEOUT=10000
ExecStart=/bin/sh c 'timeout $CEPH_VOLUME_TIMEOUT /usr/sbin/ceph-volume-systemd %i'
--

This make sense, under some special situation, the disk might not be ready for a long time, eg. waiting for vault to unseal if the disk was encrypted by vault.
But ceph-volume-systemd might give up too soon, By default it will only try about 30 (CEPH_VOLUME_SYSTEMD_TRIES) * 5 (CEPH_VOLUME_SYSTEMD_INTERVAL) seconds. and this may also confuse the ceph user as they may expect ceph-volume to keep trying for about 10000 seconds.

Actions
Copy link
 #2

Updated by Jan Fajerski over 4 years ago

I'm not sure setting a 10k second default time-out makes sense. If this makes sense for some use cases, setting the env should be a good compromise.
Going with this default would mean users with a broken disk need to wait over 2.5h to find out about that.

Actions
Copy link
 #3

Updated by Jan Fajerski over 4 years ago

Apologies, I see now the what you mean. Makes sense to me.

Actions
Copy link

Also available in: Atom PDF