[MDEV-27820] MariaDB crash if replication present between 2 different galera clusters Created: 2022-02-11  Updated: 2023-04-11  Resolved: 2023-04-11

Status: Closed
Project: MariaDB Server
Component/s: Galera, Platform Debian, Replication, SSL
Affects Version/s: 10.5.13, 10.5.14
Fix Version/s: N/A

Type: Bug Priority: Major
Reporter: Goburdhun Sarvesh Sharma Assignee: Jan Lindström (Inactive)
Resolution: Fixed Votes: 0
Labels: 10.5, debian11, galera_4, replication
Environment:
  • OS: Debian 11 \
  • Sys/Kernel info: Linux theflashbox 5.10.0-11-amd64 #1 SMP Debian 5.10.92-1 (2022-01-18) x86_64 GNU/Linux \
  • Type: Physical server \
  • Tested on MariaDB versions: 10.5.13 and 10.5.14 \
  • Galera version: galera-4 : 26.4.9, 26.4.10 and 26.4.11
  • Hardware specifications:
  • raid1 - nvme, 4.0T
  • Memory: 128G, CPU(s): 32 (2 threads per core, 16 cores per socket, 1 socket)
  • Model name: AMD Ryzen 9 5950X 16-Core Processor
  • Architecture: x86_64

Attachments: Text File error.log_bug.txt    

 Description   

Current infrastructure:
1. Galera cluster (GC1) running MariaDB 10.3.32
2. Galera cluster (GC2) running MariaDB 10.5.13 (tested also on 10.5.14)
3. Asynchronous replication setup from GC1 -> GC2
4. Both clusters have SSL configured
5. Active traffic on GC1 (on node1 only)

Working Scenario:
1. GC2 running with 3-cluster members: up and running
2. Replication configured between GC1 (node1) -> GC2 (node1): up and running

Crash:
1. On GC2 (node1), stop mariadb, and start -> Crash
2. Repeats the same messages on each start attempt

Workaround:

  • Add skip_replica_start in configuration file and start mariadb: OK
  • Then start replica threads (io_thread, sql_thread): Replication up and running
    > Note: Most of the time also fixes "Node dropped from cluster" error message.

Crash Report: (attached as error.log_bug.txt)

Can it be related to the mariadb_upgrade? (https://jira.mariadb.org/browse/MDEV-27789)

Thanking you for checking.

 Comments   
Comment by Goburdhun Sarvesh Sharma [ 2023-03-03 ]

Fixed in 10.5.16
Can be closed now.

Generated at Thu Feb 08 09:55:50 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.