* John ([EMAIL PROTECTED]) spake thusly: > > Note that 2.4 kernels die (after 14 days uptime here) with "out of > > memory". I wouldn't use a 2.4 kernel for a production machine until > > that is fixed. > > Ive never had a memory problem with a 2.4 kernel. I actually ran a machine > with NO powersupply fan (it had burnt out) for about 6 months before getting > a new one, with absolutely no downtime. It was kinda nice in the winter > actually, kept my feet warm :)
Yeah, well, that doesn't mean I've imagined it. Out of curiosity, datestamp on http://www.kernel.org/pub/linux/kernel/v2.4/linux-2.4.0.tar.bz2 is 04-Jan-2001. Is it already 4th of July where you are? (4 Jan + 6 months = ?) I did some more googling (at http://groups.google.com) and found that bug still exists in 2.4.5-ac3. Apparently the system runs out of swap after X days of uptime (dep. on the load/job mix etc; X = 13 here) because VM doesn't release swap space. According to Alan Cox it's a known bug in all 2.4 kernels. The bug is hard to catch because it may or may not manifest itself, depending on your VM configuration and job mix. Bottom line: it's more likely to hit production servers that stay up 24/7. You'll never see it on a dual-boot home box because swap is cleared every time you reboot to GameOS, besides, a home box doesn't run too many jobs. I haven't seen it until 2.4.5 because new kernel releases came out (& I had to reboot the box after installing new kernel) before the bug had a chance to bit. That's nothing new: 2.2 kernels didn't stabilize until 2.2.16 or so IIRC. Dima -- E-mail dmaziuk at bmrb dot wisc dot edu (@work) or at crosswinds dot net (@home) http://www.bmrb.wisc.edu/descript/gpgkey.dmaziuk.ascii -- GnuPG 1.0.4 public key We're sysadmins. Sanity happens to other people. -- Chris King in asr
