[MDEV-19731] 10.1.36-MariaDB - InnoDB: Failing assertion: !table->data_dir_path - Server crashes every 5 min with assertion errors after enabling Monyog Disk info Collections Created: 2019-06-11  Updated: 2019-06-18

Status: Open
Project: MariaDB Server
Component/s: Storage Engine - InnoDB
Affects Version/s: 10.1.36
Fix Version/s: 10.1

Type: Bug Priority: Major
Reporter: Vanessa Alcocer Assignee: Marko Mäkelä
Resolution: Unresolved Votes: 0
Labels: None
Environment:

Production

Attachments: File MONyog.log-20190519.gz     File MONyog.log-20190602.gz     File MONyog.log.05302019.gz     File MONyog.log.06082019_3.gz     File MONyog.log.06082019_disablediskaftertesting.gz     File MONyog.log_4.gz     File ch0pxdbcore06-error.log.06012019.gz     File ch0pxdbcore06-error.log.06082019.1.gz     File ch0pxdbcore06-error.log.06082019.2.gz     File ch0pxdbcore06-error.log.06082019.3.gz    

 Description   

After enabling Monyog Disk Info Collections (8.8.0 version) which enable queries against metadata tables ,to get sizes of tables, and type of tables per engine, PROD server start crashing every 5 min and corrupt innodb tables in one particular server.

According to Monyog the queries run are innofensive but after turning off the Disk info collection the crashes to the server stop.

Example
This query will return TableName and number of records it has. */ SELECT TABLE_SCHEMA AS `Database`, TABLE_NAME AS `Table`, TABLE_ROWS AS `Rows` FROM information_schema.TABLES

 Comments   
Comment by Elena Stepanova [ 2019-06-17 ]

Please paste or attach the contents of the server error log.

Comment by Vanessa Alcocer [ 2019-06-18 ]

Elena

I added the errorlog files for one particular server even that I have several servers crashing with the same symptoms. The constant crashes create some corruption in some tables that I fix. I was not aware that a select statement on metadata could crash so badly the server so I start checking for configuration changes but the problem was the monyog metadata queries. Once I disable the Disk Info collections running every 5 min crashes stop completely.

After we reboot the servers during the weekend is when we will begin to experience the crashes. It appears to me that Monyog take 40 to 2 min running the select and in some way this perhaps block metadata tables and a crash occur. This server is close to 500GB in size and has large tables.

1) We want to understand how these queries (select) can create an assertion issue.
2) What's an assertion error?

Generated at Thu Feb 08 08:53:55 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.