DB errors with 100% disk space utilized

Permalink
I wired my house with Cat6 the other day and moved my servers into their own space. During the transition, I forgot to plug in my external drive used for backup. Unfortunately, my Ubuntu Server backed up to the external drive directory and without the drive attached and filled up the partition that I used for the distribution.

As a consequence, there were some troubles with database backup... I host 3 WP sites and 2 C5 sites. Only the C5 sites returned the MySQL error of "mysqldump:Got error:145: Table './waclean/Logs' is marked as crashed and should be repaired when using Lock Tables.'"

It appears that this is a C5 inherent problem, as my WP domains did not experience these problems. The sites still functioned well. The solution was to navigate to the /var/lib/mysql/database-folder and execute 'myisamchk Logs.* -r.' This worked for both databases.

I am not complaining, rather I am attempting to determine how to eliminate this issue in the future.

rbucky
 
Remo replied on at Permalink Reply
Remo
did you check the table types?

are you using myisam for all your tables?

maria or innodb have a quite different way "to recover"
rbucky replied on at Permalink Reply
rbucky
I use MyISAM engine for all. Are you suggesting that C5 might benefit from another engine such as InnoDB?
Remo replied on at Permalink Reply
Remo
whether you use MyISAM or InnoDB is not directly related to Concrete5 as it doesn't use transactions very often. At least I haven't seen it yet.

I basically use InnoDB when I need row-level locking but otherwise MyISAM is usually faster and supports fulltext indexing.

I just asked because I thought this might have causes the different behavior. I'm having a hard time to understand why Concrete5 shouldn't survive a crash. At least I'm not seeing a relation between it. It's all about MySQL imho. But I'd have to reproduce this..