So I've been running into some serious problems with earl, my fileserver.
When I'm trying to read big batches of files at once (in particular, when copying off MP3s or doing mass resizings on image galleries), my whole system will lock up for a good five to ten minutes. The hard drive light is on solid during this time, and the system is largely unresponsive. I can usually keep the shell alive and responding, but disk access is queued up until the problem goes away.
The weird part: no errors, anywhere. Nothing in the logs. I have yet to connect a monitor to watch, but I doubt the OS is dumping anything to the console—Linux is pretty good about snagging that sort of thing and stashing it in the logs.
Originally, I suspected this was a problem with ReiserFS, but I've just tried reformatting a drive as ext2, and the problem came up when killing off the device. I'm going to remount the drive and mirror my image gallery on it, but I'm guessing the problem is going to return.
So, I'm not really sure what to do. It's pretty clearly a hardware problem, but I've got no idea what, exactly, is failing. I'll pull the sides off the case tomorrow morning to see if that helps—things get really hot in there with six drives going all the time. It could also be that the mainboard is really starting to crap out, as it's had blown caps for at least a year now.
Shazbot.
Leave a comment