Derek Suzuki
2004-Mar-02 21:15 UTC
[Ocfs-users] Odd errors while mounting an OCFS filesystem
Hello again. I am setting up a new pair of servers to run RAC. They're connected via fibre-channel to a hardware RAID array, and both are able to see the exposed LUNs. When I create an OCFS filesystem on one node with mkfs.ocfs, I can mount it. When I try to mount from the other node, however, it fails. After that, the filesystem is left in a state where neither node can mount it. The kernel throws errors such as the following: (3261) ERROR: status = -999, Common/ocfsgensysfile.c, 339 (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 71 (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 389 (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 842 (3261) ERROR: status = -999, Linux/ocfsmount.c, 292 (3261) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 ocfs: Unmounting device (8,1) on node1 (node 0) ocfs: Adding node2 (node 0) to clustered device (8,1) (3264) ERROR: Systemfile cannot allocate contiguously 8 blocks, Common/ocfsgencreate.c, 358 (3264) ERROR: status = -28, Common/ocfsgensysfile.c, 456 (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 104 (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 389 (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 842 (3264) ERROR: status = -28, Linux/ocfsmount.c, 292 (3264) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 ocfs: Unmounting device (8,1) on node1 (node 1) ocfs: Adding node2 (node 0) to clustered device (8,1) (3275) ERROR: Systemfile cannot allocate contiguously 80 blocks, Common/ocfsgencreate.c, 358 (3275) ERROR: status = -28, Common/ocfsgensysfile.c, 456 On a whim, I tried reformatting with ocfstool, and then I was able to mount the filesystem from both nodes with no problems. I figure that mkfs.ocfs may just be broken, in which case I'll just create all of my filesystems with ocfstool. I do worry, however, about the overall stability of the filesystems in the long run. The servers are running Red Hat Enterprise 3 AS with the latest (9.0.1) SMP kernel. I have the 1.0.9-12 release of OCFS. Any thoughts? Derek
Wim Coekaerts
2004-Mar-02 21:22 UTC
[Ocfs-users] Odd errors while mounting an OCFS filesystem
hm do you have ultra small volumes or so ? On Tue, Mar 02, 2004 at 07:15:32PM -0800, Derek Suzuki wrote:> Hello again. I am setting up a new pair of servers to run RAC. > They're connected via fibre-channel to a hardware RAID array, and both are > able to see the exposed LUNs. > When I create an OCFS filesystem on one node with mkfs.ocfs, I can > mount it. When I try to mount from the other node, however, it fails. > After that, the filesystem is left in a state where neither node can mount > it. The kernel throws errors such as the following: > > (3261) ERROR: status = -999, Common/ocfsgensysfile.c, 339 > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 71 > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 389 > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 842 > (3261) ERROR: status = -999, Linux/ocfsmount.c, 292 > (3261) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 > ocfs: Unmounting device (8,1) on node1 (node 0) > ocfs: Adding node2 (node 0) to clustered device (8,1) > (3264) ERROR: Systemfile cannot allocate contiguously 8 blocks, > Common/ocfsgencreate.c, 358 > (3264) ERROR: status = -28, Common/ocfsgensysfile.c, 456 > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 104 > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 389 > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 842 > (3264) ERROR: status = -28, Linux/ocfsmount.c, 292 > (3264) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 > ocfs: Unmounting device (8,1) on node1 (node 1) > ocfs: Adding node2 (node 0) to clustered device (8,1) > (3275) ERROR: Systemfile cannot allocate contiguously 80 blocks, > Common/ocfsgencreate.c, 358 > (3275) ERROR: status = -28, Common/ocfsgensysfile.c, 456 > > On a whim, I tried reformatting with ocfstool, and then I was able > to mount the filesystem from both nodes with no problems. I figure that > mkfs.ocfs may just be broken, in which case I'll just create all of my > filesystems with ocfstool. I do worry, however, about the overall stability > of the filesystems in the long run. > The servers are running Red Hat Enterprise 3 AS with the latest > (9.0.1) SMP kernel. I have the 1.0.9-12 release of OCFS. > Any thoughts? > > Derek > _______________________________________________ > Ocfs-users mailing list > Ocfs-users@oss.oracle.com > http://oss.oracle.com/mailman/listinfo/ocfs-users
Derek Suzuki
2004-Mar-02 22:15 UTC
[Ocfs-users] Odd errors while mounting an OCFS filesystem
Yes. For testing purposes we created a 14MB volume. -----Original Message----- From: Wim Coekaerts To: Derek Suzuki Cc: 'ocfs-users@oss.oracle.com' Sent: 3/2/2004 7:22 PM Subject: Re: [Ocfs-users] Odd errors while mounting an OCFS filesystem hm do you have ultra small volumes or so ? On Tue, Mar 02, 2004 at 07:15:32PM -0800, Derek Suzuki wrote:> Hello again. I am setting up a new pair of servers to run RAC. > They're connected via fibre-channel to a hardware RAID array, and bothare> able to see the exposed LUNs. > When I create an OCFS filesystem on one node with mkfs.ocfs, Ican> mount it. When I try to mount from the other node, however, it fails. > After that, the filesystem is left in a state where neither node canmount> it. The kernel throws errors such as the following: > > (3261) ERROR: status = -999, Common/ocfsgensysfile.c, 339 > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 71 > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 389 > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 842 > (3261) ERROR: status = -999, Linux/ocfsmount.c, 292 > (3261) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 > ocfs: Unmounting device (8,1) on node1 (node 0) > ocfs: Adding node2 (node 0) to clustered device (8,1) > (3264) ERROR: Systemfile cannot allocate contiguously 8 blocks, > Common/ocfsgencreate.c, 358 > (3264) ERROR: status = -28, Common/ocfsgensysfile.c, 456 > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 104 > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 389 > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 842 > (3264) ERROR: status = -28, Linux/ocfsmount.c, 292 > (3264) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 > ocfs: Unmounting device (8,1) on node1 (node 1) > ocfs: Adding node2 (node 0) to clustered device (8,1) > (3275) ERROR: Systemfile cannot allocate contiguously 80 blocks, > Common/ocfsgencreate.c, 358 > (3275) ERROR: status = -28, Common/ocfsgensysfile.c, 456 > > On a whim, I tried reformatting with ocfstool, and then I wasable> to mount the filesystem from both nodes with no problems. I figurethat> mkfs.ocfs may just be broken, in which case I'll just create all of my > filesystems with ocfstool. I do worry, however, about the overallstability> of the filesystems in the long run. > The servers are running Red Hat Enterprise 3 AS with the latest > (9.0.1) SMP kernel. I have the 1.0.9-12 release of OCFS. > Any thoughts? > > Derek > _______________________________________________ > Ocfs-users mailing list > Ocfs-users@oss.oracle.com > http://oss.oracle.com/mailman/listinfo/ocfs-users
Derek Suzuki
2004-Mar-02 22:50 UTC
[Ocfs-users] Odd errors while mounting an OCFS filesystem
Great. Thanks for the tip. -----Original Message----- From: Wim Coekaerts To: Derek Suzuki Cc: 'Wim Coekaerts '; ''ocfs-users@oss.oracle.com' ' Sent: 3/2/2004 8:21 PM Subject: Re: [Ocfs-users] Odd errors while mounting an OCFS filesystem riht doesn't work you need like a few 100mb at least 2.5 or 3mb is where first data starts when a node mounts, it creates 32 system files (or there abouts) so when a second node mounts it will create those and there just won't be enough room. I think like 20-30mb is the Absoluut minimum On Tue, Mar 02, 2004 at 08:15:39PM -0800, Derek Suzuki wrote:> Yes. For testing purposes we created a 14MB volume. > > -----Original Message----- > From: Wim Coekaerts > To: Derek Suzuki > Cc: 'ocfs-users@oss.oracle.com' > Sent: 3/2/2004 7:22 PM > Subject: Re: [Ocfs-users] Odd errors while mounting an OCFS filesystem > > hm do you have ultra small volumes or so ? > > On Tue, Mar 02, 2004 at 07:15:32PM -0800, Derek Suzuki wrote: > > Hello again. I am setting up a new pair of servers to run RAC. > > They're connected via fibre-channel to a hardware RAID array, andboth> are > > able to see the exposed LUNs. > > When I create an OCFS filesystem on one node with mkfs.ocfs, I > can > > mount it. When I try to mount from the other node, however, itfails.> > After that, the filesystem is left in a state where neither node can > mount > > it. The kernel throws errors such as the following: > > > > (3261) ERROR: status = -999, Common/ocfsgensysfile.c, 339 > > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 71 > > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 389 > > (3261) ERROR: status = -999, Common/ocfsgenmisc.c, 842 > > (3261) ERROR: status = -999, Linux/ocfsmount.c, 292 > > (3261) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 > > ocfs: Unmounting device (8,1) on node1 (node 0) > > ocfs: Adding node2 (node 0) to clustered device (8,1) > > (3264) ERROR: Systemfile cannot allocate contiguously 8 blocks, > > Common/ocfsgencreate.c, 358 > > (3264) ERROR: status = -28, Common/ocfsgensysfile.c, 456 > > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 104 > > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 389 > > (3264) ERROR: status = -28, Common/ocfsgenmisc.c, 842 > > (3264) ERROR: status = -28, Linux/ocfsmount.c, 292 > > (3264) ERROR: Error mounting volume, Linux/ocfsmain.c, 310 > > ocfs: Unmounting device (8,1) on node1 (node 1) > > ocfs: Adding node2 (node 0) to clustered device (8,1) > > (3275) ERROR: Systemfile cannot allocate contiguously 80 blocks, > > Common/ocfsgencreate.c, 358 > > (3275) ERROR: status = -28, Common/ocfsgensysfile.c, 456 > > > > On a whim, I tried reformatting with ocfstool, and then I was > able > > to mount the filesystem from both nodes with no problems. I figure > that > > mkfs.ocfs may just be broken, in which case I'll just create all ofmy> > filesystems with ocfstool. I do worry, however, about the overall > stability > > of the filesystems in the long run. > > The servers are running Red Hat Enterprise 3 AS with the latest > > (9.0.1) SMP kernel. I have the 1.0.9-12 release of OCFS. > > Any thoughts? > > > > Derek > > _______________________________________________ > > Ocfs-users mailing list > > Ocfs-users@oss.oracle.com > > http://oss.oracle.com/mailman/listinfo/ocfs-users