I am having no luck restoring from a tokudb backup. I am using tokudb version 5.7.23-24 on mysql 5.7.23-24 (centos 7.6). I’ve followed the docs on how to backup tokudb.
set tokudb_backup_dir=‘/data/backups’;
This backs up successfully. I rsync the files that the backup created in /data/backups/mysql_data_dir to the remote system under the my.cnf datadir directory and I also rsync’d the backup files in /data/backups/mysql_log_bin to the log_bin directory on the remote system.
When I go to start mysql on the remote system it is just a continuous loop of restarting. The db never recovers and I see this in the log file
Checksum failure while reading header in file ./dbname/path_to_some_file.tokudb.
I’ve tried multiple times to backup and restore and it never works.
There doesn’t seem to be much documentation on restoring tokudb. I imagine it should work like backing up and restoring an xtrabackup backup.
My ultimate goal is to setup a master slave environment.
Any pointers on the correct way to restore a tokudb backup would be most helpful.
I am also running into this [URL][PS-5193] TRACE logs on 56-5.6.42-rel84.2.el6.x86_64 (possible due to tokudb?) - Percona JIRA
I assume a database restart is needed to stop the excessive logging.
This tokudb is painful!
Hi, any luck sorting this out? Running into the same issue and wondering what we are supposed to do to get the server to start. Delete the files…? I should be able to restore the data from backups provided i can get the server started again…
So going back thru my notes, I think I noticed that /var was filling up with log messages from enabling tokudb backup.
A restart did not work as the logs were still going crazy. I tried this command but it didn’t work
sudo ps_tokudb_admin --disable-backup -u root -p
What I had to do to solve my issue was to remove this from /etc/my.cnf
[mysqld_safe]
preload-hotbackup
and then restart mysql. Hopefully this helps
Actually not sure if what I just mentioned will help you. What I have in my notes is more to the logging issue and not the restore issue.
I do see that I was never able to restore and configure mysql in a master/slave with tokudb 
Hi, thanks for the response. I suspect i have an actual issue with corrupted data and that won’t help me start the server.
I am getting errors like this:
Checksum failure while reading header in file: {filename}.tokudb
According to this presentation, checksum issues are not recoverable… so i guess its start again, and slowly restore data from backups.
I’m beginning to think TokuDB is just not worth the engineering hassle for my applications…
Cheers