Messages in this thread | | | Date | Mon, 24 Jan 2005 14:50:30 +0100 | From | Pascal Dameme <> | Subject | [nfsd-bug ?] ESTALE returned after a while when using nfsdfs (2.6.10 kernel) |
| |
Hello,
When a locally exported directory is mounted other itself using nfs V3, after a few minutes, the nfs servers starts issueing "ESTALE" on previously perfectly accessible files ...
This behavior has been observed on redhat Fedora core 2, Suse SLES 9 *and* 2.6.10 (from kernel.org) kernels. The less activity there is, the fastest the problem appears ...
For some reason, it manifests *only if the nfsdfs filesystem is mounted* (in "legacy" mode, where the filesystem is not mounted, the system behaves normally for at least a week, whereas with the filesystem mounted, ESTALE is returned after at most 30 minutes)
Herafter, you will find a test scenario to reproduce the problem, as well as all information I have dug so far .
Anyone ?
Best regards, -- Pascal Dameme.
---------------------------------------------------------------------------------------------------------------- The test scenario to reproduce the problem is as follow (the test machine is a SuSE distribution running a freshly compiled 2.6.10 kernel):
#start nfsd /etc/rc.d/nfsserver start #export test directory exportfs -o rw,insecure,no_root_squash,no_subtree_check 127.0.0.1:/test/dir #mount mount -o hard,nolock,vers=3,proto=udp 127.0.0.1:/test/dir /test/dir
The following is a trace of what happens :
atchoum:~ # while true; do date;ls -ld /test/dir; sleep 60;done Thu Jan 6 14:33:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:34:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:35:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:36:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:37:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:38:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:39:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:40:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:41:41 CET 2005 drwxr-xr-x 8 root root 360 Dec 7 13:46 /test/dir Thu Jan 6 14:42:41 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:43:41 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:44:41 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:45:42 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:46:42 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:47:42 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:48:42 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:49:42 CET 2005 /bin/ls: /test/dir: Stale NFS file handle Thu Jan 6 14:50:42 CET 2005 /bin/ls: /test/dir: Stale NFS file handle
I enabled the NFS debug messages, this is what is seen in the syslog file around the problem:
Jan 6 14:40:41 atchoum kernel: NFS: revalidating (0:f/113840) Jan 6 14:40:41 atchoum kernel: NFS call getattr Jan 6 14:40:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 14:40:41 atchoum kernel: nfsd: GETATTR(3) 12: 00000001 02000800 0001bcb0 00000000 00000000 00000000 Jan 6 14:40:41 atchoum kernel: nfsd: fh_verify(12: 00000001 02000800 0001bcb0 00000000 00000000 00000000) Jan 6 14:40:41 atchoum kernel: NFS reply getattr Jan 6 14:40:41 atchoum kernel: NFS: nfs_update_inode(0:f/113840 ct=1 info=0x6) Jan 6 14:40:41 atchoum kernel: NFS: (0:f/113840) revalidation complete Jan 6 14:41:41 atchoum kernel: NFS: revalidating (0:f/113840) Jan 6 14:41:41 atchoum kernel: NFS call getattr Jan 6 14:41:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 14:41:41 atchoum kernel: nfsd: GETATTR(3) 12: 00000001 02000800 0001bcb0 00000000 00000000 00000000 Jan 6 14:41:41 atchoum kernel: nfsd: fh_verify(12: 00000001 02000800 0001bcb0 00000000 00000000 00000000) Jan 6 14:41:41 atchoum kernel: NFS reply getattr Jan 6 14:41:41 atchoum kernel: NFS: nfs_update_inode(0:f/113840 ct=1 info=0x6) Jan 6 14:41:41 atchoum kernel: NFS: (0:f/113840) revalidation complete Jan 6 14:41:42 atchoum kernel: exp_export: export of non-dev fs without fsidfound domain localhost Jan 6 14:41:42 atchoum kernel: found fsidtype 0 Jan 6 14:41:42 atchoum kernel: found fsid length 8 Jan 6 14:41:42 atchoum kernel: Path seems to be <> Jan 6 14:42:41 atchoum kernel: NFS: revalidating (0:f/113840) Jan 6 14:42:41 atchoum kernel: NFS call getattr Jan 6 14:42:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 14:42:41 atchoum kernel: nfsd: GETATTR(3) 12: 00000001 02000800 0001bcb0 00000000 00000000 00000000 Jan 6 14:42:41 atchoum kernel: nfsd: fh_verify(12: 00000001 02000800 0001bcb0 00000000 00000000 00000000) Jan 6 14:42:41 atchoum kernel: NFS reply getattr Jan 6 14:42:41 atchoum kernel: nfs_revalidate_inode: (0:f/113840) getattr failed, error=-116 Jan 6 14:43:41 atchoum kernel: NFS: revalidating (0:f/113840) Jan 6 14:43:41 atchoum kernel: NFS call getattr Jan 6 14:43:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 14:43:41 atchoum kernel: nfsd: GETATTR(3) 12: 00000001 02000800 0001bcb0 00000000 00000000 00000000 Jan 6 14:43:41 atchoum kernel: nfsd: fh_verify(12: 00000001 02000800 0001bcb0 00000000 00000000 00000000) Jan 6 14:43:41 atchoum kernel: NFS reply getattr Jan 6 14:43:41 atchoum kernel: nfs_revalidate_inode: (0:f/113840) getattr failed, error=-116
Somehow, it seems that check_export gets confused ...
I tried to mount the directory using the fsid= option, this seems to help a little, but after some time, the following message appears in the syslog:
Jan 6 18:35:41 atchoum kernel: NFS: nfs_update_inode(0:f/113904 ct=1 info=0x6) Jan 6 18:35:41 atchoum kernel: NFS: (0:f/113904) revalidation complete Jan 6 18:36:41 atchoum kernel: NFS: revalidating (0:f/113904) Jan 6 18:36:41 atchoum kernel: NFS call getattr Jan 6 18:36:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 18:36:41 atchoum kernel: nfsd: GETATTR(3) 8: 00010001 00000309 00000000 00000000 00000000 00000000 Jan 6 18:36:41 atchoum kernel: nfsd: fh_verify(8: 00010001 00000309 00000000 00000000 00000000 00000000) Jan 6 18:36:41 atchoum kernel: NFS reply getattr Jan 6 18:36:41 atchoum kernel: NFS: nfs_update_inode(0:f/113904 ct=1 info=0x6) Jan 6 18:36:41 atchoum kernel: NFS: (0:f/113904) revalidation complete Jan 6 18:37:41 atchoum kernel: NFS: revalidating (0:f/113904) Jan 6 18:37:41 atchoum kernel: NFS call getattr Jan 6 18:37:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 18:37:41 atchoum kernel: nfsd: GETATTR(3) 8: 00010001 00000309 00000000 00000000 00000000 00000000 Jan 6 18:37:41 atchoum kernel: nfsd: fh_verify(8: 00010001 00000309 00000000 00000000 00000000 00000000) Jan 6 18:37:41 atchoum kernel: nfsd: Dropping request due to malloc failure! Jan 6 18:37:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 18:37:41 atchoum kernel: nfsd: GETATTR(3) 8: 00010001 00000309 00000000 00000000 00000000 00000000 Jan 6 18:37:41 atchoum kernel: nfsd: fh_verify(8: 00010001 00000309 00000000 00000000 00000000 00000000) Jan 6 18:37:41 atchoum kernel: nfsd: Dropping request due to malloc failure! Jan 6 18:37:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 18:37:41 atchoum kernel: nfsd: GETATTR(3) 8: 00010001 00000309 00000000 00000000 00000000 00000000 Jan 6 18:37:41 atchoum kernel: nfsd: fh_verify(8: 00010001 00000309 00000000 00000000 00000000 00000000) Jan 6 18:37:41 atchoum kernel: nfsd: Dropping request due to malloc failure! Jan 6 18:37:41 atchoum kernel: nfsd_dispatch: vers 3 proc 1 Jan 6 18:37:41 atchoum kernel: nfsd: GETATTR(3) 8: 00010001 00000309 00000000 00000000 00000000 00000000 Jan 6 18:37:41 atchoum kernel: nfsd: fh_verify(8: 00010001 00000309 00000000 00000000 00000000 00000000) Jan 6 18:37:41 atchoum kernel: nfsd: Dropping request due to malloc failure!
Looks like a memory leak ...
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |