MODx site down

Here’s an odd one I’m looking for ideas on.

My site went down for at least a half-hour the other night. Several users tried from different machines and found the site dead.

During that time, (I think), a series of error messages were recorded in the MODx system event log, all of which are parse errors referring to the Maxigallery database calls to get the single random photo to put on the home page.

All say: "Execution of a query to the database failed - Got error 28 from storage engine "


I was gone when it happened so I’m trying to piece things together. I think this happened during a bulk email session using phpList. phpList uses a different database than the MODx database but I guess it’s possible that the bulk email taxed things to the point that a timeout occurred for the MaxiGallery calls, although error 28 is supposed to indicate a lack of storage space on the drive.

phpList sent less than 500 emails but they all had a PDF attachment (I’ve begged my users to just put a link to the PDF file on the web site instead of using an attachment but I can’t seem to get them to listen). I have phpList set to time the messages so as not to choke the server but maybe the PDF file is taking more time than my delay.

Any thoughts appreciated...

Bob

Sounds to me like they killed the database during the sending session. Is it a shared server on top of that?

Quote from: rthrash at Jul 17, 2007, 08:54 PM

Sounds to me like they killed the database during the sending session. Is it a shared server on top of that?

It’s a shared server and the /home dir was 94% full last time I checked. I suspected the /tmp directory but it’s only 3%. I wonder if phpList spools everthing to disk before sending. The PDF file was about 500k and was sent to almost 400 addresses.

The phpList send is throttled down so it shouldn’t really threaten the DB and it’s a separate DB from the MODx one.

The worst of it is that I can’t really test anything until next month when the next newsletter goes out.


Bob