Bug #103926 | tablespace is missing for table innodb_undo_001 | ||
---|---|---|---|
Submitted: | 7 Jun 2021 12:58 | Modified: | 30 Jun 2021 9:59 |
Reporter: | Ting-Lin Chu | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Enterprise Backup | Severity: | S2 (Serious) |
Version: | 8.0.22 | OS: | Red Hat |
Assigned to: | MySQL Verification Team | CPU Architecture: | x86 |
[7 Jun 2021 12:58]
Ting-Lin Chu
[7 Jun 2021 21:16]
MySQL Verification Team
Hi, What is your settings for innodb_data_home_dir ? I'm not reproducing this with blank innodb_data_home_dir. Thanks
[8 Jun 2021 2:53]
Ting-Lin Chu
mysql> show variables like 'innodb_data_home_dir'; +----------------------+-------+ | Variable_name | Value | +----------------------+-------+ | innodb_data_home_dir | | +----------------------+-------+ 1 row in set (0.00 sec) We don't have set innodb_data_home_dir
[8 Jun 2021 9:26]
MySQL Verification Team
Hi, I will need more information as with just this I can not reproduce the problem. - what setup you have (is this group replication / innodb cluster setup or, master-slave setup or .. since you mention "read node") - how exactly are you doing the backup? What is the exact command line you use with MySQL enterprise backup - do you experience the same issue with mysqldump or getting dump from MySQL shell ? Thanks
[8 Jun 2021 10:36]
Ting-Lin Chu
- what setup you have (is this group replication / innodb cluster setup or, master-slave setup or .. since you mention "read node") Innodb cluster with 3 nodes. mysql enterprise backup on read node 1 mysqldump backup on read node 2 Both read nodes have missing undo tablespace information. but MySQL are working fine. No error logs. - how exactly are you doing the backup? What is the exact command line you use with MySQL enterprise backup ON Read node 1: $ mysqlbackup --defaults-file=/etc/my.cnf --backup-image=/backup/tp-db-02-data.mbi_`date +%d%m_%H%M` --backup- dir=/backup --compress --no-history-logging --user=root -p backup-to-image $ mysqlbackup --defaults-file=/etc/my.cnf --incremental=optimistic --start-lsn=3494881944146 --backup-dir=/backup//incr1 -- backup-image=/backup/tp-db-02-data-incrl.mbi_`date +%d%m_%H%M` backup-to-image On read node 2: /usr/bin/mysqldump -uroot -p --skip-lock-tables --order-by-primary --single- transaction --quick --master-data=2 --routines --triggers - do you experience the same issue with mysqldump or getting dump from MySQL shell ? I experienced the same issue with the read node that run mysqldump, but I don't have using MySQL shell. mysqldump command also work as expected.
[23 Jun 2021 8:52]
MySQL Verification Team
Hi, I did not manage to reproduce this. Is this a system that was originally installed ad 8.0.22 or was it upgraded from 5.x or some earlier 8.0 ?
[24 Jun 2021 9:37]
Ting-Lin Chu
This system that is originally installed mysql-commercial-server-8.0.22-1.1.el7.x86_64 and mysql-commercial-backup-8.0.22-1.1.el7.x86_64. I will attaching undo_1_trunc.log and undo_2_trunc.log . But they seem like are binary files.
[24 Jun 2021 9:38]
Ting-Lin Chu
undo 1 trunc log
Attachment: undo_1_trunc.log (application/octet-stream, text), 16.00 KiB.
[24 Jun 2021 9:38]
Ting-Lin Chu
undo 2 trunc log
Attachment: undo_2_trunc.log (application/octet-stream, text), 16.00 KiB.
[30 Jun 2021 9:59]
MySQL Verification Team
Hi, This is a duplicate of 32654667 and is fixed in 8.0.26 Thanks Bogdan