[MCOL-2077] InnoDB select table failed with 'table not in engine' from User Module 2 Created: 2019-01-10  Updated: 2023-10-26  Resolved: 2020-03-27

Status: Closed
Project: MariaDB ColumnStore
Component/s: ?
Affects Version/s: 1.2.2
Fix Version/s: 1.2.6

Type: Bug Priority: Minor
Reporter: David Hill (Inactive) Assignee: Todd Stoffel (Inactive)
Resolution: Not a Bug Votes: 1
Labels: documentation
Environment:

Amazon AMI 2um/3pm system



 Description   

Customer reporting error from the table select on Innodb tables from UM2, when they worked from UM1.

MariaDB [XXXXXX]> select count(1) from XXXXXXX; ERROR
1932 (42S02): Table 'XXXXXX' doesn't exist in engine

Problem turned out that the InnoDB table structure file 'ibdata1' wasnt replicated from UM1 to UM2 when the rsync command is run, which is part of the upgrade process and when 'enablemysqlreplication' is run from the mcsadmin console.

Need to investigate why this table isnt replicated from um1 to other nodes via rsync.

---------------------------
work around : shutdown system and scp 'ibdata1' table from um1 to um2. startsystem



 Comments   
Comment by David Hill (Inactive) [ 2019-06-06 ]

The problem is the rysch.sh scripts isnt copying the file ibdata1 to the slave nodes

Comment by Andrew Hutchings (Inactive) [ 2019-06-13 ]

It shouldn't rsync those files from or to a running system. So many bad things will happen, scping can cause issues as well depending on how things are configured. mariabackup should be used manually in this specific scenario. This is something to document.

Generated at Thu Feb 08 02:33:33 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.