|
Apache logging stops at 4k boundaries
I'm running Apache/1.3.27 on a Slackware install (2.4.18 kernel, SMP, SCSI harddrives... it's an IBM rackmount server), and it's been working fine for quite a while. In fact I've had almost no problems with it. However I was trying to look at my access logs recently and I realized that it hadn't logged anything since June 4th. (Yes, I checked my system date :P ) Anyway, I tried rotating the logs, because I figured maybe it was too big for apache (I hadn't rotated the logs in quite a while, they were huge.) It wouldn't log to the new clean access logfile either, even after a full system reboot. I tried adding stuff to the logfile manually, and lo and behold, when I added a newline it would magically start logging again. I thought it was fixed, but it suddenly stopped again, when the logfile was exactly 4096 bytes (the size of one block on my FS.) I added another newline and it started working again, up until 8192. Tried it again, and it worked up to 12288 bytes. Apparently Apache just refuses to increase the size of the logfile past a block boundary. I doubt it's a problem with the disk, filesystem, etc. because I can write to the file perfectly otherwise. All permissions are perfect too. Does anyone know if this is a well-known problem? I couldn only find one reference to something that sounded like it, and it didn't have any answers. (It's in the linuxmanagers mailing list archives, I can't post the URL here because I'm a new user.)
So, is there a reason/fix for this? (Perhaps an upgrade?) Or am I going to have to pipe it through a logger daemon to circumvent this?
|
