[MDEV-10243] Systemd does not perform --wsrep-recover on restart Created: 2016-06-16  Updated: 2016-06-16  Resolved: 2016-06-16

Status: Closed
Project: MariaDB Server
Component/s: Galera, Packaging
Affects Version/s: 10.1.14
Fix Version/s: 10.1.15

Type: Bug Priority: Major
Reporter: Philip Stoev (Inactive) Assignee: Nirbhay Choubey (Inactive)
Resolution: Duplicate Votes: 0
Labels: None

Issue Links:
Duplicate
is duplicated by MDEV-10004 Galera's pc.recovery process fails in... Closed

 Description   

The systemd scripts packaged with MariaDB execute mysqld directly. This in turn prevents the --wsrep-recover and --wsrep-start-position mechanism from kicking in, as this mechanism is only found in mysqld_safe.

As a result, restarting a crashed mysqld via systemd will always cause SST to take place even if IST is available:

Jun 16 05:53:23 centos7 mysqld[3091]: 2016-06-16  5:53:23 140118583347328 [Note] WSREP: Assign initial position for certification: -1, protocol version: -1
...
Jun 16 05:53:23 centos7 mysqld[3091]: 2016-06-16  5:53:23 140118583032576 [Warning] WSREP: Failed to prepare for incremental state transfer: Local state seqno is undefined: 1 (Operation not permitte
Jun 16 05:53:23 centos7 mysqld[3091]: at galera/src/replicator_str.cpp:prepare_for_IST():491. IST will be unavailable.



 Comments   
Comment by Elena Stepanova [ 2016-06-16 ]

It might have been fixed in MDEV-10004. Assigning to nirbhay_c to confirm.

Comment by Philip Stoev (Inactive) [ 2016-06-16 ]

Yes, this seems to have fixed it. Apologies for the duplicate, feel free to close it, as I can not.

Generated at Thu Feb 08 07:40:44 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.