Innobackupex should maintain a pid file
Bug #1086286 reported by
Raghavendra D Prabhu
This bug affects 5 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Percona XtraBackup moved to https://jira.percona.com/projects/PXB |
Triaged
|
Medium
|
Unassigned | ||
2.0 |
Won't Fix
|
Medium
|
Unassigned | ||
2.1 |
Triaged
|
Medium
|
Unassigned | ||
2.2 |
Triaged
|
Medium
|
Unassigned | ||
2.3 |
Triaged
|
Medium
|
Unassigned |
Bug Description
In light of the bug report here https:/
It is better if innobackupex takes a --pid option in which it
writes its pid.
It serves dual purpose:
1. Check if innobackupex started by a wrapper is still running
(in presence of say multiple mysql daemons and/or multiple
innobackupex running, pidof is not feasible.)
2. If innobackupex dies a non-graceful death, presence of pid
file (which should be removed towards the end in the script)
indicates that.
tags: | added: pxc |
To post a comment you must log in.
Yes, it's necessary for XtraBackup to detect existing backup, and maybe a "lock" file is more accurate for this situation.
For our environment, there are many MySQL instances and we backup daily. Some instances are too huge to finish backup in 24h, so there maybe two running backup processes, and this will failing backup, for example the 'xtrabackup_ suspended_ 2' file of this first run backup was removed by the second.
And the lock file should be specific to one instance, such as put in backup tmpdir or take a name unique to an instance(ip+port), cause there could be more than one mysqld instance in a host.