Ahhh, I had a Howard Rogers experience, aka website problems. Friday night, I couldn’t login on my WordPress site. What ever I did, on mySQL database level or else, I wasn’t allowed to login anymore and got a nice message telling me that I used an incorrect password.
First thing that can to my mind was, of course, that the site had been hacked. So I made a backup of all my files on disk and created database backups of my mySQL databases. I afterwards checked my providers support side, speedxs.nl if I had missed something. The only thing I could find was that a message in the maintenance section saying:
Maintenance webserver Damia
As you may have noticed, there has been discovered an instability in the webserver damia. Therefore maintenance will be scheduled sooner than planned.
The maintenance process will be done next Friday and this is expected to be done during a time period of 2 hours maximum.
That my problems and the scheduled maintenance took place during the same time period, looked suspicious, so I created a phpinfo.php file to check the environment. The outcome of the report mentioned that PHP had been rebuild on Friday afternoon at approx. 15:00 hours. What did they do?
With this info in mind, I copied all files back to the webserver (apparently it didn’t look like a hack attempt). I tried now to reset my MD5 password hash as demonstrated on the “WordPress phpMyAdmin Guide”. Strangely enough it didn’t help; also the “forgot password” routine didn’t solve my problem. I still wasn’t allowed to login with a newly generated password.
Being frustrated, I send an email, asking what they had done on Friday. After an e-mail or 2, 3, they reported that they completely overhauled the environment; with other words also upgraded mySQL from version 4 to version 5 (from 4.0.x via 4.1. to 5.0.44).
While doing their whole maintenance thing, they also (extra surprise) limited me on memory, so my site showed now memory problems:
Fatal error: Allowed memory size of 8388608 bytes exhausted (tried to allocate 266802 bytes)
On a website I found an easy test regarding the memory problems. Renaming the plugin directory, and therefore disabling them, would temporary solve these memory problems.
Their explanation regarding the mySQL issue:
The upgrade has been done because of “end of life” reasons.
OK, but why not tell this in the first place? I did a lot of tests in the meantime, including installing a test environment with the latest WordPress distro… As always, good, honest communication, definitely helps and it is something I can live with…
Long story short. At the end the mySQL problem was solved by them by “restoring” the mySQL databases (file based? via an import? I still don’t know); the memory problems were solved, by granting me more memory.
Ah, lets face it. I now have correct backups in place again, upgraded all domains to the latest WordPress distributions and know a little bit more about problems regarding mySQL upgrades. Sarcastic…?