[MDEV-25739] Failed to create a new provider '/usr/lib/galera/libgalera_smm.so' with options Created: 2021-05-20 Updated: 2021-08-17 Resolved: 2021-07-19 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | Galera |
| Affects Version/s: | 10.5.9, 10.5.10, 10.5 |
| Fix Version/s: | 10.4.21, 10.5.12, 10.6.4 |
| Type: | Bug | Priority: | Blocker |
| Reporter: | Jaroslav | Assignee: | Alexey |
| Resolution: | Fixed | Votes: | 2 |
| Labels: | None | ||
| Issue Links: |
|
||||||||||||
| Description |
|
After upgrade to 10.5.10 and also on 10.5.9 we started to see following issue where node will never sync back to cluster and end in failed state
On the other nodes we also could see
There is nothing we can do just to delete the disk for the node and let it fully resync. From linked tickets it seems the /var/lib/mysql/galera.cache got corrupted somehow and deleting it "solves" the issue. |
| Comments |
| Comment by Sergio Charrua [ 2021-05-28 ] |
|
same issue here with 10.5.9, but with only 1 node (no Galera nodes running at the time) |
| Comment by Alexey [ 2021-07-19 ] |
|
Gcache ring buffer is not synced to disk and is not guaranteed to be consistent on restart. Hence some bogus numbers can be read from disk and care must be taken of potential exceptions. In particular here insert may throw exception for bogus seqno value. Fixed in commit 4bb58377cf3ee02e4c69ce329c2e099b07c79368 |
| Comment by Alexey [ 2021-07-19 ] |
|
Fixed in Galera commit 4bb58377cf3ee02e4c69ce329c2e099b07c79368 |