Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.11.9
-
None
-
Rocky Linux 9.4
10.11.9-1.el9.x86_64
Description
I am running a huge database of about 3TB of data with MariaDB 10.11.9 on a mirrored ZFS vdev with SSDs.
Whenever I try to use Mariabackup with this database, it fails near the end with a message like:
mariabackup --target-dir=/mnt/ --slave-info --backup
...
[00] FATAL ERROR: [timestamp] Was only able to copy log from 148582842890306 to 148638022288286, not 148676138598341; try increasing innodb_log_file_size
The innodb_log_file_size is 32GB; I even tried 128GB, but it does not fix the problem.
Some time after Mariabackup starts, it repeatedly complains until the end with the same message:
Retrying read of log at LSN=148638022288286
The backuptarget has plenty of space free.
I increased innodb_log_file_size and tried this on three different slave-servers (all handle about 1TB to 3TB data) with the very same result. After a few hours of mariabackup-runtime, the backup fails with this error message.
The expectation is, that the backup should finishes successfully.
Attachments
Issue Links
- is blocked by
-
MDEV-34062 mariadb-backup --backup is extremely slow at copying ib_logfile0
- Closed
- relates to
-
MDEV-23947 reflink support for mariabackup
- Open