Bug 425421 - gfs mount attempt hangs if no more journals available
gfs mount attempt hangs if no more journals available
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: gfs-kmod (Show other bugs)
All Linux
low Severity low
: ---
: ---
Assigned To: Robert Peterson
GFS Bugs
: 460233 (view as bug list)
Depends On:
Blocks: 475312
  Show dependency treegraph
Reported: 2007-12-14 15:41 EST by Corey Marthaler
Modified: 2010-01-11 22:28 EST (History)
0 users

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2009-01-20 16:18:33 EST
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)
Patch to fix the problem (473 bytes, patch)
2008-02-01 15:19 EST, Robert Peterson
no flags Details | Diff

  None (edit)
Description Corey Marthaler 2007-12-14 15:41:58 EST
Description of problem:
I created a gfs filesystem on a cluster with three nodes.

[root@grant-02 ~]# gfs_mkfs -O -j 2 -p lock_dlm -t GRANT-CLUSTER:gfs2
Device:                    /dev/grant/gfs2
Blocksize:                 4096
Filesystem Size:           19593744
Journals:                  2
Resource Groups:           300
Locking Protocol:          lock_dlm
Lock Table:                GRANT-CLUSTER:gfs2

All Done

When I attempt mount on the third node, it hangs.

Trying to join cluster "lock_dlm", "GRANT-CLUSTER:gfs2"
Joined cluster. Now mounting FS...
GFS: fsid=GRANT-CLUSTER:gfs2.2: can't mount journal #2
GFS: fsid=GRANT-CLUSTER:gfs2.2: there are only 2 journals (0 - 1)
Dec 14 14:34:57 grant-03 kernel: GFS 0.1.19-7.el5_1.1 (built Nov 12 2007 19:27:d
Dec 14 14:34:57 grant-03 kernel: Trying to join cluster "lock_dlm", "GRANT-CLUS"
Dec 14 14:34:57 grant-03 kernel: Joined cluster. Now mounting FS...
Dec 14 14:34:57 grant-03 kernel: GFS: fsid=GRANT-CLUSTER:gfs2.2: can't mount jo2
Dec 14 14:34:57 grant-03 kernel: GFS: fsid=GRANT-CLUSTER:gfs2.2: there are only)

[root@grant-03 ~]# strace mount /dev/grant/gfs2 /mnt/gfs2
execve("/bin/mount", ["mount", "/dev/grant/gfs2", "/mnt/gfs2"], [/* 21 vars */]) = 0
brk(0)                                  = 0xee55000


stat("/dev/grant/gfs2", {st_mode=S_IFBLK|0660, st_rdev=makedev(253, 3), ...}) = 0
rt_sigprocmask(SIG_BLOCK, ~[TRAP SEGV RTMIN RT_1], NULL, 8) = 0
open("/dev/grant/gfs2", O_RDONLY)       = 3
fstat(3, {st_mode=S_IFBLK|0660, st_rdev=makedev(253, 3), ...}) = 0
lseek(3, 0, SEEK_SET)                   = 0
read(3, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...,
69632) = 69632
close(3)                                = 0
stat("/sbin/mount.gfs", {st_mode=S_IFREG|0755, st_size=42000, ...}) = 0
child_tidptr=0x2aaaaaaba540) = 3262

Version-Release number of selected component (if applicable):
[root@grant-03 ~]# rpm -qa | grep gfs

How reproducible:
Comment 1 Nate Straz 2007-12-19 15:03:51 EST
Planning on removing GFS-kernel.  Moving all bugs to gfs-kmod.
Comment 2 Robert Peterson 2008-02-01 15:19:53 EST
Created attachment 293764 [details]
Patch to fix the problem

This was an easy one.  When Wendy did the fast statfs patch, she
forgot to deallocate the statfs (formerly license) file on errors.
Comment 3 Robert Peterson 2008-02-01 15:25:16 EST
Setting flags for inclusion in RHEL5.  It's already fixed in RHEL4.
Comment 4 Robert Peterson 2008-02-01 15:32:08 EST
Looks like Wendy tried to remedy this problem with a commit to the RHEL5
branch of CVS but she got it wrong.  Her comment referenced bug #231904.
Here is the diff that shows the put against the wrong inode, the quota inode
rather than the license inode.


So I guess we should still fix this.  Again, it looks okay in RHEL4.
Comment 5 Robert Peterson 2008-04-14 16:44:06 EDT
This patch was tested on the roth-0{1,2,3} cluster at RHEL5.
I pushed the patch to the master, STABLE2 and RHEL5 branches of
the cluster git tree.  Changing status to modified.
Comment 6 Corey Marthaler 2008-08-27 14:58:28 EDT
*** Bug 460233 has been marked as a duplicate of this bug. ***
Comment 7 Corey Marthaler 2008-08-27 15:39:49 EDT
Is there any way we could have a better error message?

[root@taft-03 tmp]# mount /dev/taft/mirror1 /mnt/taft1
/sbin/mount.gfs: error mounting /dev/mapper/taft-mirror1 on /mnt/taft1: Invalid

Granted, the info is in the log...
Aug 27 14:32:11 taft-03 kernel: Trying to join cluster "lock_dlm", "TAFT:1"
Aug 27 14:32:11 taft-03 kernel: Joined cluster. Now mounting FS...
Aug 27 14:32:11 taft-03 kernel: GFS: fsid=TAFT:1.3: can't mount journal #3
Aug 27 14:32:11 taft-03 kernel: GFS: fsid=TAFT:1.3: there are only 3 journals
(0 - 2)

...but "Invalid argument" makes me think I misspelled something or that the device is bad or failed. If not, then this bug can be marked verified.
Comment 9 Nate Straz 2008-12-08 15:25:01 EST
Verified with gfs-utils-0.1.18-1.el5 and kernel-2.6.18-124.el5.
Comment 11 errata-xmlrpc 2009-01-20 16:18:33 EST
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.


Note You need to log in before you can comment on or make changes to this bug.