I'm running MW 1.25 and PHP 5.5.33 with mysql 5.5.
I'm having a problem with one of my locally hosted wikis. I got an email from a user that our site was down, and discovered that the backend mysql database had been overwhelmed with processes giving
LCStoreDB::finishWrite
as part of their info when I use show processlist; from the mysql prompt. There were also some very long-running processes like this
COMMIT /* LoadBalancer::commitMasterChanges 66.249.64.57 */
I closed access to the wiki by putting a die(); message at the top of LocalSettings.php. The number of mysql processes dropped, but the db was still not responsive. I then made what was probably a mistake by trying to kill processes in mysql. From reading up on why the killed processes are still there, it seems that killing a commit causes a rollback, which can be very slow... 3-4 hours later, it's still not cleared up. Now I see a bunch of processes that look like
INSERT /* MessageBlobStore::insertMessageBlob 66.249.75.103 */ IGNORE INTO `msg_resource` (mr_lang,m
From what I've read, it sounds like I have to just wait for the rollback to complete. But while I'm waiting... is there anything I should be doing with my configuration to avoid this happening in the future?
Thanks,
Jim Hu
=====================================
Jim Hu
Professor
Dept. of Biochemistry and Biophysics
2128 TAMU
Texas A&M Univ.
College Station, TX 77843-2128
979-862-4054