On Thursday 09 October 2003 10:29, Mihai RUSU wrote: > On Thu, 9 Oct 2003, Mihai RUSU wrote: > > Hi again > > > > Because my 2.4.20 + ctx16 kernel crashes very 2-6 hours I need a stable > > solution ASAP. I am compiling right now 2.4.22 + ctx17a as beeing the > > current stable acording to the site:
Hi! I am experiencing hangs aswell, but I don't know if they are related with vserver at all. I'm not sure if you mean crashes as in kernel panics or if it just freezez. In my case it freezez completly after about 2 days. So for now I have a reboot crontab running every midnight (which isn't nice). I am running with the latest kernel (2.4.23pre6) and almost the latest ctx patch (17e?). But I have patched in evms, quota support for reiserfs, vroot and grsecurity. It seems to me that it has something to do with access the filesystem (because it tends to crash when updatedb is running, after turning that crontab off it crashes at random.). So first I was thinking this has to be the reiserfs quota patch because I had to change it to resolv patching rejects. But I ripped that part of the kernel and still the same results. Actually, first I had the rmap patch in and thought that might be it, but that didn't solve it either (ripping it out). Because it hangs and doesn't give me any errors I'm reading up on how to debug kernel problems, have gotten as far as enabling the sysrq. but, still don't know what causes it.. I'm running 3 vservers, got 2 gb of ram and 2 gb of swap.. it seems that the vservers are just using more and more and more and more memory, except I don't know for what (according to vserver-stat that is, don't know how exact that is.. ) and the inode_cache has crazy values (compared to any other linux system I run). Oh, and sometimes the load goes up like crazy (especially right before it hangs, don't know why though). Have been running vserver completly stable with all the same patches (except vroot, and other kenel 2.4.21 and ctx patch). And it has always been rock solid, so it came to me as quite a surprise when it didn't work out this time. :) Anyways, tell me more about your crash, maybe there is information to be derived from this. -- -- Asbjorn Sannes [EMAIL PROTECTED] www.sannes.org
