Recently, I noticed I was having issues with the Zimbra logger. As a result, I found this wiki page.
So I issued a REPAIR TABLE statement, and after 5+ hours wait, I decided to investigate, and found I'd run into the exact same issue as outlined in this MySQL bug report. My du/df returns exactly the same results as outlined in that bug (an empty and yet full /tmp/ directory)
Doing a tail servername.com.err returns several lines similar to the below..
Any thoughts? Right now, I'm EXTREMELY loath to abort the "REPAIR TABLE" process, yet I am pretty sure it's stuck...
071219 15:56:04 [ERROR] /opt/zimbra/logger/mysql/libexec/mysqld: Disk is full writing '/tmp/STYAnvMQ' (Errcode: 28). Waiting for someone to free space... Retry in 60 secs