cinder/releasenotes/notes/feature-multi-process-backup-8cf5ad5a0cf9b2d5.yaml
Gorka Eguileor 373b524041 Support multiple processes on Cinder Backup
Cinder Backup has always been run as a single process, and this works
fine if the number of backups is small, but there are deployments where
a big number of backups are run simultaneously, and in such cases we
will find a bottleneck on CPU intensive operations, like compression and
SHA calculations, since we are using a single process and therefore a
single CPU core.

This patch adds support to define the number of processes we want to run
on the backup service using a new "backup_processes" configuration
option.

When running multiple processes they will all be running as children of
a parent process, but when running a single process it will run on its
own.

To preven race conditions and avoid the backup service from becoming
available before it should only the first process will do the initial
cleanup and all the others will wait on a lock until it is released by
this first process.

Change-Id: Ib43095024754a6219eb51cc0663913fac10bb642
2018-03-14 10:52:25 +01:00

8 lines
318 B
YAML

---
features:
- |
Cinder backup now supports running multiple processes to make the most of
the available CPU cores. Performance gains will be significant when
running multiple concurrent backups/restores with compression. The number
of processes is set with `backup_workers` configuration option.