On Mon, Mar 07, 2005 at 06:54:52AM -0700, Kevin Pendleton wrote: > Hi all, > > I'm new to this vserver thing, I'm taking up where someone else left > off, so I'm sure this is a newbie question, but your help is greatly > appreciated. > > I have a dual 1.1 GHz server with 1 GB of RAM running ~40 vservers.
well, congratulation, that's quite something with 1GB (probably on a celeron server too ... ) > The server has started crashing lately on a pretty regular basis > with a ton of VM: killing process httpd (different processes) > --alloc_pages: 0- order allocation failed (gfp=0xld2/0) written > to the console. I have to hard boot the server to bring it back up. yep that means that the last piece of low memory was used up by some kernel stuff ... > The server is running software RAID on three SCSI disks and is running > 2.4.27-vs1.29-rc2 on Red Hat 9. It didn't seem to happen before one of > the SCSI drives died and was rebuilt. > > My question is this: which is the right solution here? add more memory > or upgrade the kernel/vserver? Does this combination have a known > memory leak or is it more likely that there is a memory leak with one of > the apps running in a particular vserver? how much swap is available there? do you use strict no overcommit or is it running in overcommit mode? > I'm sure the answer is probably both, that I need more memory and I need > to upgrade and include other features like cpu throttling - is there > memory throttling? there are memory limits and cpu throttling in 1.9.x (the 2.6 branch) best, Herbert > Thanks in advance for your advice/help. > > Kevin > > > _______________________________________________ > Vserver mailing list > [email protected] > http://list.linux-vserver.org/mailman/listinfo/vserver _______________________________________________ Vserver mailing list [email protected] http://list.linux-vserver.org/mailman/listinfo/vserver
