Messages in this thread | | | Date | Tue, 21 Sep 1999 09:27:34 -0400 | From | Wade Hampton <> | Subject | Re: problems with RedHat 6.0 and NFS |
| |
Dave Meyer wrote: > > We have an NFS server running RedHat 6.0 (running 2.2.5-22 UP) that > sometimes behaves really badly under heavy loads. What happens is > that dmesg will show a bunch of lines like[snip] > and the nfsd processes will come to a complete halt. In addition, > they can't be stopped - `kill -9' means nothing to them. The > filesystem can't be unmounted, and after rebooting there are > invariably problems that fsck must repair.
I am having similar lockups with 2.2.12, 2.2.13pre[8,9], etc. The messages I typically get are:
EXT2-fs error (device sd(8,33): ext2_new_block: Free blocks count corrupted for block group 235 [only appears on the Dell sometimes]
and when NFS locks up I get:
nfs_dentry_delete: //xfer: ino=<some number>, count=2, nlink=2
About all I can do is get a dump with the SYSRQ keys. NFS seems to be locked up in fs/nfs/inode.c: nfs_free_dentries() > This problem seems to crop up most often when several people are > compiling on the same remote filesystem. My problem crops up when I am NFS copying 1G of files from from box A to box B and 1G from box B to box A, all via 100-T. Then I execute the following on either box (e.g., box A): on box A: cd <some dir on NFS mounted directory box B> mkdir testfiles cp <some files> testfiles in a loop, do: ln -s testfiles testfiles-1 ls testfiles* rm testfiles-1 mv testfiles testfiles-1 ln -s testfiles-1 testfiles ls testfiles* rm testfiles mv testfiles-1 testfiles enddo
System: Penguin dual PIII, RedHat 6.0, Dell WS 400 dual PII/300, RedHat 6.0
both are SMP boxes with 2.2.12, 2.2.13pre8, 2.2.13pre9, and 2.2.13pre9+ Andrea's patches (almost pre10) including ikd.
I have been working with Alan Cox and Andrea Arcangeli on this issue.
If anyone has any similar lockups or other info, please share it with us....
Cheers, -- W. Wade, Hampton <whampton@staffnet.com>
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |