| Summary: | mount.nfs (net-fs/nfs-utils-1.2.3-r1) generates page allocation failure | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Gerhard Hintermayer <gerhard.hintermayer> |
| Component: | [OLD] Core system | Assignee: | Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel> |
| Status: | RESOLVED NEEDINFO | ||
| Severity: | critical | ||
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | Output from dmesg | ||
|
Description
Gerhard Hintermayer
2011-05-10 12:10:25 UTC
Created attachment 272725 [details]
Output from dmesg
the output of dmesg
the corresponding line in /etc/fstab is: aprol1:/srv/backup /srv/backup nfs auto 0 0 the command for mounting was: mount /srv/backup any advice on how to proceed on this (production) server is appreciated. I'm running a webserver and a postgresql server on it, should I reboot, to be in a clean memory state or can I simply ignore the error message ? Can you tell us the version of the last working kernel? Did you do a kernel upgrade recently (and/or nfs-utils)? (In reply to comment #4) > Can you tell us the version of the last working kernel? Did you do a kernel > upgrade recently (and/or nfs-utils)? I did both a kernel update and an rebuild of nfs-utils on the weekend of may 1st. The kernel I had before was 2.6.36-gentoo-r8, the nfs-utils were the same before, the reinstall was due to some broken dependancies. I don't know if the error was'nt already in 2.6.36-gentoo-r8, because I discovered that automountig the nfs share at startup did not work since I rebooted the machine in march for upgrading to 2.6.36-gentoo-r8. Ok, but you don't get the page alloc error when booting a 2.6.36 kernel, right? Btw, can you post the output of cat /proc/sys/vm/min_free_kbytes ? If it's less than 10M(10k kbytes actually), can you set it to something like 16M or 64M? Probably this isn't a critical bug. I think that the kernel tries to do a somewhat big allocation (order=4) and the VM subsystem temporarily can't find enough available memory to allocate. Thus you get the page alloc error, and then when you remount, it works. Try setting min_free_kbytes to 16M/64M, and hopefully it'll fix your problem. (In reply to comment #6) > Ok, but you don't get the page alloc error when booting a 2.6.36 kernel, right? I don't know, since automounting the nfs dir did not work at the last server reboot using. I have to change the fstab entry to using the IP of the remote host, since nfsmount script lacks a "need dns" :-( > Btw, can you post the output of cat /proc/sys/vm/min_free_kbytes ? > If it's less than 10M(10k kbytes actually), can you set it to something like > 16M or 64M? > 3798 > Probably this isn't a critical bug. I think that the kernel tries to do a > somewhat big allocation (order=4) and the VM subsystem temporarily can't find > enough available memory to allocate. Thus you get the page alloc error, and > then when you remount, it works. > > Try setting min_free_kbytes to 16M/64M, and hopefully it'll fix your problem. I'll try increasing min_free_kbytes. Thanks. Any updates to report here? |
