Details
-
Bug
-
Status: Stalled (View Workflow)
-
Critical
-
Resolution: Unresolved
-
10.6.14
-
None
Description
When there's a wrong variable name MariaDB Server stops and exits.
There is thou a misleading error message logged by Galera which can lead to think that SST failed because of that (even when no SST actually happened.
 |
2023-06-21 15:04:07 3 [Note] WSREP: Server status change joiner -> initializing
|
2023-06-21 15:04:07 3 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
|
2023-06-21 15:04:07 0 [Note] InnoDB: Compressed tables use zlib 1.2.13
|
2023-06-21 15:04:07 0 [Note] InnoDB: Number of pools: 1
|
2023-06-21 15:04:07 0 [Note] InnoDB: Using crc32 + pclmulqdq instructions
|
2023-06-21 15:04:07 0 [Note] InnoDB: Using Linux native AIO
|
2023-06-21 15:04:07 0 [Note] InnoDB: Initializing buffer pool, total size = 134217728, chunk size = 134217728
|
2023-06-21 15:04:07 0 [Note] InnoDB: Completed initialization of buffer pool
|
2023-06-21 15:04:07 0 [Note] InnoDB: 128 rollback segments are active.
|
2023-06-21 15:04:07 0 [Note] InnoDB: Creating shared tablespace for temporary tables
|
2023-06-21 15:04:07 0 [Note] InnoDB: Setting file './ibtmp1' size to 12 MB. Physically writing the file full; Please wait ...
|
2023-06-21 15:04:07 0 [Note] InnoDB: File './ibtmp1' size is now 12 MB.
|
2023-06-21 15:04:07 0 [Note] InnoDB: 10.6.14 started; log sequence number 3807484; transaction id 185
|
2023-06-21 15:04:07 0 [Note] Plugin 'FEEDBACK' is disabled.
|
2023-06-21 15:04:07 0 [Note] InnoDB: Loading buffer pool(s) from /run/media/claudio/FedoraData/myharem/instances/mariadb-enterprise-10.6.14-9-centos-9-x86_64.30614/data/ib_buffer_pool
|
2023-06-21 15:04:07 0 [ERROR] /run/media/claudio/FedoraData/myharem/instances/mariadb-enterprise-10.6.14-9-centos-9-x86_64.30614/bin/mariadbd: unknown variable 'table_definitio_cache=4K'
|
2023-06-21 15:04:07 0 [ERROR] Aborting
|
2023-06-21 15:04:07 3 [ERROR] WSREP: sst_received failed: State wait was interrupted
|
2023-06-21 15:04:07 2 [ERROR] WSREP: Application received wrong state:
|
|
And this is without a wrong variable name:
2023-06-21 15:01:13 3 [Note] WSREP: Server status change joiner -> initializing
|
2023-06-21 15:01:13 3 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
|
2023-06-21 15:01:13 0 [Note] InnoDB: Compressed tables use zlib 1.2.13
|
2023-06-21 15:01:13 0 [Note] InnoDB: Number of pools: 1
|
2023-06-21 15:01:13 0 [Note] InnoDB: Using crc32 + pclmulqdq instructions
|
2023-06-21 15:01:13 0 [Note] InnoDB: Using Linux native AIO
|
2023-06-21 15:01:13 0 [Note] InnoDB: Initializing buffer pool, total size = 134217728, chunk size = 134217728
|
2023-06-21 15:01:13 0 [Note] InnoDB: Completed initialization of buffer pool
|
2023-06-21 15:01:13 0 [Note] InnoDB: 128 rollback segments are active.
|
2023-06-21 15:01:13 0 [Note] InnoDB: Creating shared tablespace for temporary tables
|
2023-06-21 15:01:13 0 [Note] InnoDB: Setting file './ibtmp1' size to 12 MB. Physically writing the file full; Please wait ...
|
2023-06-21 15:01:13 0 [Note] InnoDB: File './ibtmp1' size is now 12 MB.
|
2023-06-21 15:01:13 0 [Note] InnoDB: 10.6.14 started; log sequence number 3803750; transaction id 172
|
2023-06-21 15:01:13 0 [Note] Plugin 'FEEDBACK' is disabled.
|
2023-06-21 15:01:13 0 [Note] InnoDB: Loading buffer pool(s) from /run/media/claudio/FedoraData/myharem/instances/mariadb-enterprise-10.6.14-9-centos-9-x86_64.30614/data/ib_buffer_pool
|
2023-06-21 15:01:13 0 [Note] InnoDB: Buffer pool(s) load completed at 230621 15:01:13
|
2023-06-21 15:01:13 0 [Note] Server socket created on IP: '0.0.0.0'.
|
2023-06-21 15:01:13 0 [Note] Server socket created on IP: '::'.
|
2023-06-21 15:01:13 0 [Warning] WSREP: Ignoring server id for non bootstrap node.
|
2023-06-21 15:01:13 0 [Note] WSREP: wsrep_init_schema_and_SR 0x0
|
2023-06-21 15:01:13 0 [Note] WSREP: Server initialized
|
2023-06-21 15:01:13 0 [Note] WSREP: Server status change initializing -> initialized
|
2023-06-21 15:01:13 0 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
|
2023-06-21 15:01:13 3 [Note] WSREP: Recovered position from storage: fc65b47a-0ebd-11ee-a691-eb1618faaa0e:72
|
2023-06-21 15:01:13 3 [Note] WSREP: Recovered view from SST:
|
|
You can see the difference after: [Note] InnoDB: Loading buffer pool(s)
My impression is that Galera inherits somehow an error happening in that phase wrongly attributing it to SST phase which is formally still open.
The real problem is that at next restart the node will request a full SST even if that's not really required.
This happens also with no activity whatsovere in the cluster.
To reproduce:
- Start Galera cluster
- Stop a Node
- Add non_existing_variable=ON in the config
- Start the node
- See how it fails and notice the SST failed messages
- Remove the wrong variable
- Start the node
- Observe a full SST happening