Red Hat Bugzilla – Bug 144054
page allocation failure nfsd & xfs
Last modified: 2015-01-04 17:14:41 EST
Description of problem:
There are "page allocation failure"s and the box is generally unstable.
Intel P4 2.8GHz, 2 gigs RAM
3Ware 9000 RAID card (12x200 gigs in RAID 50, for 1.7TB)
NFS server only
XFS filesystem for RAID partition
Fedora Core 2
Gigabit ethernet (e1000)
Version-Release number of selected component (if applicable):
I have crashed the box by doing a ~3 gig mysqldump via NFS while
simultaneously downloading via ftp a 4 gig ISO. This is via gigabit
The debian boxes that NFS mount it also generate lots of:
nfs: server 10.1.1.1 not responding, still trying
nfs: server 10.1.1.1 OK
Steps to Reproduce:
1. Throw loads of data at the box, particularily via NFS to the XFS
2. Just wait awhile...
Box OOPs, has "page allocation failures", just croaks outright,
becomes sluggish, etc.
Be really fast & stable.
Created attachment 109314 [details]
page allocation failures
XFS is trying to do a really huge allocation. Little surprise that it
falls over under load I guess.
XFS isn't a 'supported' filesystem of Fedora, so your best bet is to
tell the SGI folks who maintain it about this problem, and whatever
they fix we'll pick up the next time we rebase.
From the maintainers file...
Bugzilla'd w/ SGI
"The problem is 4K stack which is default in fedora and 8K in vanilla
kernel. I don't have time to make too much tests, so I just use
vanilla kernels on machines with XFS."
Fedora Core 2 has now reached end of life, and no further updates will be
provided by Red Hat. The Fedora legacy project will be producing further kernel
updates for security problems only.
If this bug has not been fixed in the latest Fedora Core 2 update kernel, please
try to reproduce it under Fedora Core 3, and reopen if necessary, changing the
product version accordingly.