What you are missing is a shared (virtual) disk.
If you believe that a disk is so shared, then format it on one
guest and then issue the following command on both.
$ debugfs.ocfs2 -R "stats" /dev/sdX | grep UUID
You should get the same UUID on both guests.
Once this works, then attempt to mount
If the hang you see is after a node (with a mounted ocfs2 volume) dies,
then it is a known one. This specific recovery bug was introduced in 1.2.7
and fixed in 1.2.8-2. 1.2.8-SLES-r3074 maps to 1.2.8-1. The fixed one should
be version r3080 or more.
If so, upgrade to the latest SLES10 SP1 kernel.
Any errors in the /var/log/messages?
Any busy locks:
http://oss.oracle.com/~smushran/.debug/scripts/scanlocks
$ scanlocks
If so, dump them on all nodes using:
$ echo R
List of domains can be gotten from this:
http://oss.oracle.com/~smushran/.debug/scripts/listdomains
Alexandre Racine wrote:
>
Alexandre Racine wrote:
> Excellent, that works great! Now that I have the locks and the domain
> name what should I do to unlock them? (Or fix the problem).
>
The locking and unlocking is handled by the dlm. I'm working
on updating the wikis with more information on debugging such
issues. For t
AFAIK:
a. There is no force umount in Linux.
b. There is no way to know whether a local fs is mounted on another node.
Luis Freitas wrote:
> Alexandra,
>
>You could use only CRS and ext3 instead of ocfs2 for this kind of
> use. You would need to register a script to force umount the
> filesy
Lazy umount is not the same as forced umount. The processes
that have active descriptors will keep reading and writing to
the fs. The processes are not killed.
Luis Freitas wrote:
> Hmm,
>
> There is a "lazy" umount:
>
>
>-l Lazy unmount. Detach the filesystem from the filesystem
>
http://oss.oracle.com/projects/ocfs2/dist/documentation/ocfs2-new-features.html
Added in 2.6.22
It will be available for enterprise kernels with ocfs2 1.4.
1.4 has just started shipping with sles10 sp2. It will be
available for (rh)el5 u2 in a months time.
Matthew Barr wrote:
> Does ocfs2 support
That means o2dlm is not the cause for the process hang.
Next, run ps:
$ ps -e -o pid,stat,comm,wchan=WIDE-WCHAN-COLUMN
Run it 6 times in 10 sec interval. This should tell us the processes
in D state and the location in kernel. Hopefully.
The next step is to to get the stack trace. alt-sysrq-t. T
; jfsCommit jfs_lazycommit
> 332 S< jfsSync jfs_sync
> 333 S< xfslogd/0 worker_thread
> 334 S< xfslogd/1 worker_thread
> 335 S< xfslogd/2 worker_thread
> 336 S< xfslogd/3 worker_thread
> 337 S< xfsda
Stop o2cb and switch node number in /etc/cluster/ocfs2.conf.
After changing on boh, restart o2cb on both.
[EMAIL PROTECTED] wrote:
>
> Hi Sunil,
>
> my lotus notes choked on the table from excel... So the two nodes have
> the following nodenumbers:
> Node ocfs2 crs/css
> byaz05 0 2
> byaz10 1 1
>
What version/kernel are you running?
We are about to release 1.2.9 that addresses one known case of o2net
consuming 100% cpu. Due out next week.
Michael S. Moody wrote:
>
> I had an instance today on several servers where the load average
> soared, and all of my apache processes were in uninterr
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=c824c3c723f2e37a00b3b739a55b28de595fd72e
Michael Moody wrote:
> Ocfs2-tools version 1.3.9
> Kernel 2.6.24-gentoo-r8
>
> Michael
>
> Sunil Mushran wrote:
>
>> What version/kernel are you
The one I am aware of is o2net spinning at 100% before
the node is fenced due to cluster timeout.
Michael Moody wrote:
> What ramifications could this bug have?
>
> Random filesystem lockups, file locks not being released?
>
> Michael
>
> Sunil Mushran wrote:
>
>
All,
We are pleased to announce the release of OCFS2 1.2.9-1 for RHEL4 and RHEL5
on x86, x86_64, ppc64 and ia64 architectures.
This release includes bug fixes, most of which have been backported from the
mainline kernel. Some of the more interesting ones have been described in
detail. For the ful
http://oss.oracle.com/bugzilla/show_bug.cgi?id=919
Fixed in 1.2.9-1. SUSE has the fix checked into their
tree. It should be out soon with sles10 sp1.
Charlie Sharkey wrote:
>
> On a two node cluster I got a reboot (core dump) on the first node.
> The /var/log/messages doesn't show anything wron
Yes. I am assuming this was from the time you had
to force shutdown the servers. If so, ignore.
Alexandre Racine wrote:
> Hi all,
>
> I had a drive freeze and after shutting down all servers and starting
> them one by one I had this in the logs. What does this tells you?
> Network problems? Thanks
The only relevant message is the first one that indicates
that srv2 has not heard from srv1 for 30 secs. The rest of the
messages are because the link broke and are more informative
than errors.
Alexandre Racine wrote:
> Well this was today, but yes before I had to force shutdown the
> machines.
>
The ocfs2 1.x packages are only meant for sles and (rh)el and
not other distributions. 2.6.18 and 2.6.16 shipped with (rh)el
and sles respectively are not the same as the original mainline
tree. So building it will require some development work.
One option for you is to ping the kernel vendor and
File a bugzilla and attach the coredump.
One thing to try would be reducing the number of slots.
[EMAIL PROTECTED] wrote:
> Dear Srs,
>
> I get this error when running "mkfs.ocfs2":
>
> =
> # mkfs.ocfs2 -b 4K -C 32K -
http://oss.oracle.com/projects/ocfs2/news/article_18.html
This is oss bugzilla#919 that has been fixed in 1.2.9-1.
Saranya Sivakumar wrote:
> Hi,
>
> We recently upgraded ocfs2 to 1.2.8 from 1.2.3 on our 4 node RAC
> production systems.
>
> On one of the nodes, we notice the following in the log
mar
>
>
> ----- Original Message
> From: Sunil Mushran <[EMAIL PROTECTED]>
> To: Saranya Sivakumar <[EMAIL PROTECTED]>
> Cc: ocfs2-users@oss.oracle.com
> Sent: Wednesday, June 18, 2008 3:54:31 PM
> Subject: Re: [Ocfs2-users] ocfs2 1.2.8 issues
>
> http://
OCFS2 is only available for the Linux kernel.
Diane Petersen wrote:
> Hello,
>
> I'm building an 11g RAC 2-node cluster on Solaris 10 (64-bit SPARC).
> Is there an OCFS2 version available and where would I find it?
> I've looked on the main oss.oracle.com/projects/ocfs2 and it's not
> obvious whe
All,
This is a heads up only for users exporting OCFS2 volumes as NFS mounts.
If not, please disregard this email.
Recently there was a bugzilla filed that mentioned observing file system
lockups when accessing OCFS2 exported volumes with FreeBSD NFS clients.
The lockups were not observed by him
The fs in 2.6.21 still uses the old very short cluster timeouts.
In mainline, the defaults were updated in 2.6.25-ish.
The faq has the details on setting them manually.
http://oss.oracle.com/projects/ocfs2/dist/documentation/ocfs2_faq.html#TIMEOUT
Carlos Xavier wrote:
> Dear Srs.
>
> I have be
g for help on compiling the kernel module, so as we
> can have a updated one for the kernel 2.6.21.5 and 2.6.24.5 distributed with
> Slackware 12.0 and 12.1.
>
> Tanks,
> Carlos Xavier.
>
> - Original Message -
> From: "Sunil Mushran" <[EMAIL PROT
Run fsck to repair that inode.
fsck.ocfs2 -f /dev/sdd1
Also, better if you upgrade the fs to 1.2.9-1.
--- Begin Message ---
I get the following error when trying to mount: Nothing changed (in any
case not that I know of). I get the same error from both nodes. Please
assist.
# mount /u02
moun
Sure.
You can mark the volume as local (man tunefs.ocfs2 or mkfs.ocfs2)
and mount it without the cluster stack (like any local file system).
You can use the ro mount option to mount it readonly.
Combine the two and you get what you want.
BTW, if the fs image on a physical ro media, the fs autom
Kruyt [EMAIL
PROTECTED]
Sent: Wednesday, June 25, 2008 8:41 AM
To: Sunil Mushran
Cc: ocfs2-users@oss.oracle.com
Subject: Re: [Ocfs2-users] mount readonly without lockmanager
Thanks,
But the filesystem is on a SAN, and shared arcoss 3 nodes. One of that node I
dont wat a lockmanager and the system
$ dd if=/dev/sdd1 of=/tmp/out count=100 bs=4K
Can you file a bugzilla and attach the /tmp/out.
Looks like it is unable to read the inode because it's
signature if off. The above command will dump the first 400K
of the device.
I want to see what the extent of the corruption is. If it is
localized
Fedora ships ocfs2 fs modules natively. You don't have to do all this.
What is missing is the tools rpm.
But the good news is that that should be available any day now
literally speaking.
Tina Soles wrote:
>
> Hello,
>
> I’m brand new to RAC and ocfs2. I need to install ocfs2, but there is
>
ply. Can you be more specific and give me the exact
> name of the native Fedora 9 rpm(s) that I need for ocfs2 and ocfs2-tools?
> Thanks.
> ----
> *From:* Sunil Mushran [mailto:[EMAIL PROTECTED]
> *Sent:* Fri 6/27/
ROTECTED]
Sent: Sunday, June 29, 2008 10:16 PM
To: Tina Soles
Cc: Sunil Mushran; ocfs2-users@oss.oracle.com
Subject: Re: [Ocfs2-users] Problems building ocfs2 rpm on Fedora 9
Hi Tina,
Sorry, I am a not RAC expert. So you may have to wait for
Sunil's
suggestion for it.
If you
Could be due to bugzilla#919 as explained in the list of fixes in 1.2.9-1.
http://oss.oracle.com/projects/ocfs2/news/article_18.html
Gabriele Di Giambelardini wrote:
> Hi, this is my output on all the 5 servers
>
> Module "configfs": Loaded
> Filesystem "configfs": Mounted
> Module "ocfs2_nodemana
mount(2) man page lists the following reasons for it to return an EBUSY:
EBUSY
source is already mounted. Or, it cannot be remounted read-only,
because it still
holds files open for writing. Or, it cannot be mounted on target
because target is
still busy (it is the working directory of some
My recommendation is for you to use an enterprise kernel... (rh)el or sles.
For shared storage, use iscsi. sles10 ships with a good iscsi target.
Firewire as a shared disk was useful when there was no inexpensive
shared disk available. That is no longer the case.
Tina Soles wrote:
> Sunil,
>
> I a
Upgrade to OCFS2 1.2.9-1 shipping with the latest SLES9 SP4 kernel
(2.6.5-7.312).
http://download.novell.com/Download?buildid=27kCZ1qWwWo~
You are most likely hitting bug#6680001 as mentioned here.
http://oss.oracle.com/projects/ocfs2/news/article_17.html
Also, you might want to tone down the he
Which kernel?
uname -a?
block/cluster sizes?
debugfs.ocfs2 -R "stats" /dev/sdX
How many nodes in your cluster?
memory?
cat /proc/meminfo
Michael Moody wrote:
>
> I use rsync to take backups of my ocfs2 filesystems (since nothing
> else really supports it out of the box). Unfortunately, it’s ve
34359738367 kB
> VmallocUsed:293184 kB
> VmallocChunk: 34359444779 kB
> HugePages_Total: 0
> HugePages_Free: 0
> HugePages_Rsvd: 0
> HugePages_Surp: 0
> Hugepagesize: 2048 kB
>
> 5 nodes in the cluster.
>
> Michael
>
> -Ori
What distro, kernel, packges versions, etc?
http://oss.oracle.com/projects/ocfs2/dist/documentation/ocfs2_faq.html#DOWNLOAD
Check the requirements for the console. You could be missing a package that
enables propagate config.
[EMAIL PROTECTED] wrote:
> I've been attempting to follow the instruct
Gabriele Di Giambelardini wrote:
> Hi to all, some time ago, I read that the ocfs have a limit for the
> subfolder. Is it possible this whren this limit gone exceeded the
> ocfs2 have those problem???
>
>
> Or some boby know the limit number?
http://oss.oracle.com/projects/ocfs2/dist/documentat
y have hundreds of thousands files
> inside the concurrent output and log directories.
>
> Regards,
> Luis
>
> --- On *Thu, 7/3/08, Sunil Mushran /<[EMAIL PROTECTED]>/* wrote:
>
> From: Sunil Mushran <[EMAIL PROTECTED]>
> Subject: Re: [Ocfs2-users] o
Post the exact command and the error message.
--- Begin Message ---
Hello,
I would like to install OCFS2 on CentOS5.2 with the 2 2.6.18-92.1.6.el5xen
kernel. When I try to compile the OCFS2 source for this kernel I see a
message nothing todo for rhel5 in the vendor section.
What I did was the fol
1.2.5 is year+ old. Suggest you upgrade to 1.2.9.
The oops is bizzare to say the least. I notice you are using
xenU kernel. 4 nodes are VMs? Just trying to understand the
layout.
Is it reproducible? Definitely upgrade to 1.2.9. If the issue
reproduces, file a bugzilla with all the details.
This
./configure --with-src=/usr/src/kernels/2.6.18-92.1.6.el5-x86_64
make
--- Begin Message ---
> Post the exact command and the error message.
This is a selection of the output, I have attached the complete output.
[EMAIL PROTECTED] ocfs2-1.3.9]# ./configure
checking build system type...
In mainline, that issue was resolved in 2.6.21. We have patches for
2.6.20 but not older than that.
Sunil
--- Begin Message ---
Hi all,
I just started with OCFS2 and set up a 2-node cluster where one node is
writing and both read from the clustered volume. Currently I'm moving
data to the volum
All,
This is to inform all SLES users that the latest SLES9 SP4 and
SLES10 SP1 kernel erratas includes OCFS2 1.2.9-1. Users running the
older kernel with 1.2.8-1 are urged to upgrade to the current release.
For more information on the changes in OCFS2, please refer to the email
announcing 1.2.9-1
In mainline, the issue was addressed in 2.6.21.
In enterprise kernels, the issue was addressed in 1.2.4-2.
If you are on (RH)EL4 or (RH)EL5, install 1.2.9-1.
If you are on SLES9 SP4 or SLES10 SP1, upgrade to the
latest kernel.
Sunil
--- Begin Message ---
From the User's Guide:
> Oracle databas
RAC is only supported on (RH)EL and SLES.
It may work with other distros, but support is a different beast.
datavolume mount option is not in mainline kernel. Oracle 10g onwards,
the database itself does not require it... as one can set filesystemio_options
to directio (init.ora param). That away
[EMAIL PROTECTED] wrote:
> By raw are you meaning raw device access without a filesystem like ocfs2
> on the volume for the voting disk? Or am I not following?
>
Raw means specifying the block device directly. So make two
partitions, say, sdd1 and sdd2, and feed that (/dev/sdd1, etc)
to the to
Email me the following info:
$ debugfs.ocfs2 -R "stats" /dev/sdX <== replace with ocfs2 device
$ stat /mnt/user/small/11/11wa1.jpg
$ stat /data/user/small/11/11wa1.jpg
Markus Meyer wrote:
> Hi all,
>
> I stumbled over a curious thing. The Linux tools "df" and "du" aren't
> working cor
Markus Meyer wrote:
> Block Size Bits: 12 Cluster Size Bits: 16
> Links: 0 Clusters: 6707596
So you have a 4TB volume. Correct? Appears mkfs chose 64K as the
cluster size. This means the smallest data allocation would be 64K.
>File: `/mnt/user/small/11/11wa1.jpg'
>Size
If you are still on 1.2.8-2, then it is a known issue fixed in 1.2.9-1.
Gabriele Di Giambelardini wrote:
> Hi to all, watching the log by more attention and in the moment when a
> node go down, I have this imformation by the kernel about o2net :
>
> Jul 10 16:52:02 be1 kernel: BUG: soft lockup -
File a bugzilla with the logs of all the machines. /var/log/messages.
Meanwhile do schedule an upgrade to 1.2.9-1. We have one fix relating
to o2net fencing that could have been in play here. But I'll need to read
the full logs to be sure.
Sunil
--- Begin Message ---
Hello,
We have a four-node R
That's 175 million files. I hope they are spread out across many
directories.
Our inodes are blocksized.
4k blocksize means 700G of metadata.
2K means 350G.
1K means 175G.
AFAIK, XFS has 256 byte inodes.
Maybe try 1K blocksize and 8K clustersize.
You would be an ideal candidate for the inlined
Sabuj Pattanayek wrote:
> Hi,
>
> I'm using OCFS2 from 2.6.26 with some patches I made that allow for
> the creation of a volume greater than 16TB:
>
> http://oss.oracle.com/pipermail/ocfs2-devel/2008-July/002568.html
> http://oss.oracle.com/pipermail/ocfs2-tools-devel/2008-July/000857.html
>
> The
That is strange.
Next time double check the mounts with:
$ cat /proc/mounts
The mount command prints the entries in /etc/mtab while
the /proc/mounts dumps the information from the kernel.
If those threads are there, it means the volume is still
mounted. Two in this case.
The entries in mtab are
Did you monitor /proc/mounts as I had suggested.
-Original Message-
>From Mark Schloss <[EMAIL PROTECTED]>
Sent Mon 7/21/2008 9:22 PM
To Sunil Mushran <[EMAIL PROTECTED]>
Cc ocfs2-users@oss.oracle.com
Subject Re: [Ocfs2-users] OCFS processes active after a umount [SEC=UNO
Please file a SR with Oracle support. Database issues are best
resolved in that forum.
__
>From Ed Gulakowski <[EMAIL PROTECTED]>
Sent Wed 7/23/2008 7:48 PM
To ocfs2-users@oss.oracle.com
Subject [Ocfs2-users] ORA-19870 and ORA-19502 During RMAN
To start off, you are using a very old version of the fs/tools.
2+ year old. Upgrading will take care of the -R option.
However, the basic problem you are experiencing will remain.
As in, ocfs2 uses blocksized inodes. You can reduce the blocksize,
but that will result in a loss of thruput as the i
rious stages in the test
> outlined below. Also, the -n option is not used on the mount.
>
> Regards
>
> Mark Schloss
>
>
> Mark Schloss | Oracle DBA | Information Technology | x0013
>
> -Original Message-
> From: Sunil Mushran [mailto:[EMAIL PROTECTED]
Hard for me to diagnose the issue with no logs. Maybe best
if you logged a bugzilla with novell and provided them with
all the logs.
Kuang, Howard [WHQKT] wrote:
>
> Hi, Sunil,
>
>
>
> I upgrade ocfs2 to 1.2.9.1 with the new kernel from Novell. The
> fencing problem is still existing. When one
Two inits take time.
1. Cluster group init.
2. Journal init.
Considering this is a 16TB volume being formatted with
4K/4K block/cluster sizes, means it has 127074 cluster groups
to initialize. So 127074 4K blocks to initialize. But this bit
should be somewhat similar to ext3.
Journal initializati
SLES10 SP2 is shipping OCFS2 1.4. We will releasing the
same for (RH)EL in the coming weeks.
-Original Message-
>From Haydn Cahir <[EMAIL PROTECTED]>
Sent Mon 7/28/2008 8:07 PM
To ocfs2-users@oss.oracle.com
Subject Re: [Ocfs2-users] OCFS2 and VMware ESX
Hi Mark,
Thanks for your reply. Ho
Use an enterprise kernel.
Tina Soles wrote:
> OK, another snag. Fedora 9 does not support RAW devices, so I can't
> configure the voting disk or OCR disk to be as such. Any suggestions? I
> think I'm "up a creek" here...
>
> -Original Message-
> From: [EMAIL PROTECTED] [mailto:[EMAIL PRO
The default mkfs params make 4 slots each with a 256M journal. That's 1G.
If you want them smaller, mkfs provides parameters to override the same.
Secondly, we compute based on the full device. Most other filesystems
deduct the blocks consumed by the fs on creation in their calculation.
Arnold Ma
No, the kernel is old. A year+ old.
Refer to this announcement below.
http://oss.oracle.com/pipermail/ocfs2-announce/2008-July/26.html
From the stack, it looks you are encountering the rename/extend race
that was fixed a long time ago.
http://oss.oracle.com/projects/ocfs2/news/article_14.htm
I believe 1.2.5-SLES-r2997 is the version of the fs and not the
tools. Meaning, an upgrade is required to the latest kernel
that is shipping 1.2.9.
As far as failure to mount goes, one reason could be that the
default timeout (10 secs) could be low. See if increasing to the
new default of 30 secs
http://oss.oracle.com/bugzilla/show_bug.cgi?id=838
Check out this bugzilla.
Tao Ma wrote:
> Hi,
>
> Masanari Iida wrote:
>> Hello Tao and Sunil,
> ]> My case, the symptom (ocfs2 failed to mount a volume using
>> /etc/fstab) happend when I reboot the system.
>> Even if it failed to mount (by /etc/f
ocfs2 home page on oss.oracle.com has support guides. Also, we have a wiki.
http://oss.oracle.com/osswiki/OCFS2/Debugging
The dlm debugging has been improved in sles10 sp2. We will be soon releasing
the details in ocfs2 1.4's user's guide.
Make sure you are running the latest sles10 sp1 kernel.
h
Do:
$ debugfs.ocfs2 -l DLM ENTRY EXIT allow
$ mkdir /dlm/test
$ debugfs.ocfs2 -l DLM off ENTRY EXIT deny
File a bugzilla and attach the /var/log/messages.
Charlie Sharkey wrote:
>
>
>
> I'm having a problem creating a dlm domain. The libo2dlm library
> returns a 'could not create domain' err
Yes. It is fully ondisk compatible.
Read the bit about file system compatibility here.
http://oss.oracle.com/pipermail/ocfs2-announce/2008-March/23.html
We will be releasing a more formal user's guide soon.
Paulo Rodrigues wrote:
> Hello,
>
> is it safe to unmount FS under 1.3.3 and mount un
This could suggest an on disk problem. Have you run fsck.ocfs2 recently?
fsck.ocfs2 -f /dev/sdX1
Paulo Rodrigues wrote:
> Hello,
>
> I'm on 2.6.24 with OCFS2 1.3.3 and every couple days this comes up in
> dmesg. I have to reboot the cluster machines, there's nothing else I
> can do. Stopping t
I am hoping we would not have this problem with the new
cluster stacks that are in development, cman and pacemaker.
But always good to hear about the issues being encountered
by the users. Well, not good... but you know what I mean.
Michael Moody wrote:
>
> I have a suggestion about the heartbeat
Does not look you used the force option. Or, you ran with the
file system mounted.
Umount the fs on all nodes and do:
$ fsck.ocfs2 -f /dev/dm-1
Paulo Rodrigues wrote:
> Hello Sunil,
>
> fsck says its clean:
>
> Checking OCFS2 filesystem in /dev/dm-1:
> label: /var/lib/dovecot/spool
This may not be related to ocfs2. Check the oracle
doc for the version of the database you are running.
It will have all the appropriate kernel settings.
Daniel Keisling wrote:
> Greetings,
>
> When attempting to start up a database on an OCFS2 filesystem, Oracle
> complains with the following
Please can you file a bugzilla and attach this stack trace.
Also attach the output of the following:
$ objdump -DSl /lib/modules/`uname -r`/kernel/fs/ocfs2/ocfs2.ko
>/tmp/ocfs2.out
Paulo Rodrigues wrote:
> Got the same error again today.
>
> BUG: unable to handle kernel NULL pointer dereference
Configure a netdump or netconsole server. It will catch the relevant
messages.
Derek Hazell wrote:
>
> Dear OCFS2 forum
>
> We run ocfs2 version 1.2.9-1 as an ocfs2 cluster on four Linux servers
> running RHEL 4 (kernel: 2.6.9-42.0.2.ELs)
>
> We are getting unexpected reboots of one of the L
Thanks. This looks like a new issue. Please log it in the bugzilla.
http://oss.oracle.com/bugzilla
Wessel wrote:
> Hello All,
>
> I’ve been running a 4-node OCFS2 cluster for about a month now, and recently
> I’ve had a total of 3 kernel errors on random nodes. This causes the machine
> to lock up
All,
We are pleased to announce the release of OCFS2 1.4. This release has
been available with Novell's SUSE Linux Enterprise Server (SLES10 SP2)
for some time now. Today we are announcing the release of the same for
Red Hat's and Oracle's Enterprise Linux (EL5 U2) distributions.
Before upgrading
Actually, the mount will fail. The clusterstack detects mismatches
during the handshake.
The 1.3.9 tools corresponds with the file system that comes with
the kernel. It was the best release at that time. However, software
development is a constant process. We are adding new features
and fixing bug
2.6.18 is very very old. We've fixed many may bugs since then.
Upgrade to a more recent kernel. Earliest I would say 2.6.21.
Or anything after that.
Dante Garro wrote:
>
> Hi all!
>
> I've just configured a new 2 node cluster and I found messages at
> kernel.log like the following:
>
> Aug 19 19:
What does "mounted.ocfs2 -d" return on the two nodes?
Corne Lombard wrote:
>
> I ran into a problem in section 16 (Install & Configure Oracle Cluster
> File System (OCFS2)) of the following article –
>
> http://www.oracle.com/technology/pub/articles/hunter_rac10gr2_iscsi_2.html#16.
>
> I have 2 R
ms to do submit_bio for read
> Index 8: took 120303 ms to do waiting for read completion
> *** ocfs2 is very sorry to be fencing this system by restarting ***
> Bootdata ok (command line is ro root=/dev/VolGroup_ID_12182/LogVol1
> console=ttyS0,9600n8)
>
>
> ##
No, the device names have nothing to do.
When you mount, mount.ocfs2 kicks off the heartbeat. When
other nodes see a new node heartbeating, o2net attempts to
connect to the node. That connect is necessary for the mount
to succeed.
My investigation would start with disk heartbeat.
# watch -d -n2
No, the fscat tools can only read certain unmounted file systems.
They cannot write. You can use it to copy data from ocfs
to ocfs2 on a box running the 2.6 kernel (sles9/10, el4/5).
Mehmet Can ÖNAL wrote:
>
> Hi everyone;
>
>
>
> we have a production system with 6 nodes of RAC upon ocfs file sy
Define export?
Do you want the nodes to be part of a cluster? As in, want
local fs semantics across nodes.
If so, use a shared device that can accomodate more than
2 nodes. drbd8 is great for what it does but also limits users
to 2 nodes.
Zack Gilburd wrote:
> Hi all,
>
> I have ocfs2 atop drbd8
So in 1.4, we have a much improved debugging infrastructure for
such issues. Check out the write on dlm debugging in the 1.4
user's guide in the chapter titled notes.
In short, you have correctly identified the lock resource. But we
need to go a step further and get the info from the dlm and see
a
ively, if you've rebooted the system that holds the lock
> would the others reclaim locks held and carry on as normal?
>
>Andy
>
>
>
> -Original Message-
> From: Sunil Mushran [mailto:[EMAIL PROTECTED]
> Sent: Tue 02/09/2008 05:21
> To: Andrew Phillips
>
Check out qs 80/81 in the faq.
http://oss.oracle.com/projects/ocfs2/dist/documentation/v1.2/ocfs2_faq.html#QUORUM
which version, distro, etc.?
Dante Garro wrote:
> Hi Sirs, I have a 2 nodes cluster.
>
> Seems (but I'm not sure)when I've rebooted one of nodes the other reboots
> itself too.
>
> Is
That will be so if KVM is buffering the ios. Which it must be
doing for performance reasons.
Bret Baptist wrote:
> On Friday 29 August 2008 18:38:08 Bret Baptist wrote:
>
>> On Thursday 28 August 2008 18:59:07 Sunil Mushran wrote:
>>
>>> If the VM is not see
Repeat the test. This time run the following on Node A
after you have killed Node B.
$ ps -e -o pid,stat,comm,wchan=WIDE-WCHAN-COLUMN
If we are lucky we'll get to see where that process is waiting.
Henri Cook wrote:
> Hi all,
>
> I have two nodes (A+B) running a DRBD file system (using OCFS2) on
rify it's still connected with a ping node and continue about
> its duties (it can even maintain read/write in a DRBD context)
>
> H
>
> Sunil Mushran wrote:
>> Check out qs 80/81 in the faq.
>> http://oss.oracle.com/projects/ocfs2/dist/documentation/v1.2/ocfs2_faq.h
What version of ocfs2-tools are you running?
Doron Tamir wrote:
>
> Hi all ,
>
>
>
> When I type :
>
> cat /etc/sysconfig/o2cb
>
>
>
> I see only
>
> 1. # O2CB_ENABELED: 'true' means to load the driver on boot.
> 2. O2CB_ENABLED=true
> 3.4.
That fix went into 2.6.26.
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=ffda89a3bf3b968bdc268584c6bc1da5c173cf12
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=a4a4891164d4f6f383cc17e7c90828a7ca6a1146
http://git.kernel.org/?p=linu
What version?
$ modinfo ocfs2
$ rpm -qa | grep ocfs2
$ uname -a
Sunil
Marshall, Richard wrote:
>
> Hello:
>
> We have a 3 node cluster, on one of the nodes the LAN cable was
> accidentally disconnected, and the node hard booted (i.e. power
> off/on). There were no indications from ocfs2 and o2
Which kernel?
The error message itself needs to be silenced. The OR should
be changed to an AND.
2184 if (ret != -EFAULT || ret != -ENOSPC)
2185 mlog_errno(ret);
But that just means we are treating this as a user error.
However, as the same
Enable some tracing:
$ debugfs.ocfs2 -l ENTRY EXIT INODE DISK_ALLOC SUPER FILE_IO NAMEI AIO allow
$ xm save
$ debugfs.ocfs2 -l ENTRY EXIT deny INODE DISK_ALLOC SUPER FILE_IO NAMEI
AIO off
File a bugzilla and attach the syslog.
Brett Worth wrote:
> Sunil Mushran wrote:
>
&g
One more thing.
Do:
$ strace -ff -o /tmp/save xm save
Zip up the traces and attach to bugzilla.
Lastly, is the guest hvm or pvm, 32-bit or 64-bit. From the packages
the host machine appears to be 64-bit.
Sunil Mushran wrote:
> Enable some tracing:
>
> $ debugfs.ocfs2 -l ENTRY E
The network protocol differs between 1.2 and 1.4. It won't work.
Marco Mililotti wrote:
> Hi all,
>
> is it possible/acceptable to mount and use an Ocfs2 filesystem on two
> machines running *different* software level? I.e.:
> - M1 that runs ocfs2 1.2.3-0.7, kernel drv ver: 1.2.5-SLES-r2997
> - M
Ensure the cluster.conf is the same across the cluster. If it is not,
edit and restart the cluster.
The "transport endpoint" error means that the tcpip connect failed.
It could be because of incorrect ip, firewall, or a bad cluster.conf.
The dmesg errors indicate that the cluster.conf could be mi
601 - 700 of 1424 matches
Mail list logo
