[MDEV-32414] [mariadb-galera] There is one node in a 3-node cluster, ibdata1, with a large file Created: 2023-10-10  Updated: 2023-10-10  Resolved: 2023-10-10

Status: Closed
Project: MariaDB Server
Component/s: Galera, Storage Engine - InnoDB
Affects Version/s: 10.6.8
Fix Version/s: N/A

Type: Bug Priority: Blocker
Reporter: luorui Assignee: Unassigned
Resolution: Duplicate Votes: 0
Labels: None
Environment:

centos 7 amd64

Attachments: PNG File image-2023-10-10-14-38-53-320.png     PNG File image-2023-10-10-14-39-06-909.png     PNG File image-2023-10-10-14-39-20-575.png    
Issue Links:
Duplicate
duplicates MDEV-21952 ibdata1 file size growing in MariaDB Closed

 Description   

MariaDB-Galera deployed using BitNami's helm, a multi-master cluster with one node with an abnormal ibdata1 file size

node1 is normal

node2 is normal

node3 is abnormal

Why is ibdata1 of node 3 so large and different from other nodes
I want ibdata1 to be smaller and not increase rapidly

 Comments   
Comment by Marko Mäkelä [ 2023-10-10 ]

This ticket seems to duplicate MDEV-21952. A possible explanation for the file growth could be that the performance of the purge of history is somewhat unpredictable. That could be improved in MDEV-32050. Another explanation could be the change buffer, if you have overridden the default that was changed in MDEV-27734.

Comment by luorui [ 2023-10-10 ]

Why is this the case with only one node? Shouldn't multi-master architectures all increase undo logs, and besides, if I deploy in a k8s cluster, can I synchronize the data of other nodes by deleting the data of the problematic node and restarting the pod?

Generated at Thu Feb 08 10:31:11 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.