ocfs2 supports private mmap r/w and shared mmap readonly.
Shared mmap writeable is the only piece missing. We should have that by 1.4.
Alexei_Roudnev wrote:
There was a clear answer, WHY it did not worked on OCFSv2:
- BerkleyDB and LDAP uses mmap to the files;
- OCFSv2 don't implement it (becau
ssage -
From: "Sunil Mushran" <[EMAIL PROTECTED]>
To: "Alexei_Roudnev" <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>; "Michael Wood" <[EMAIL PROTECTED]>;
Sent: Wednesday, December 06, 2006 1:47 PM
Subject: Re: [Ocfs2-users] OCFS2 and berkeley data
4o2cb.
*From:* [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] *On Behalf Of *Yuval Baruch
*Sent:* Thursday, December 07, 2006 4:06 PM
*To:* Alexei_Roudnev
*Cc:* Sunil Mushran; ocfs2-users@oss.oracle.com; [EMAIL PROTECTED]
*Subject:* Re: [Ocfs2-users] DMesg error on startup ...
thanks for you offer
Sorry for the delay.
We are investigating a problem we encountered when we tested the
patch with 2.6.19 (mainline). We are still investigating whether it is
limited to the mainline or is it something that can occur on the
supported kernels. (It just could be that we were lucky in our testing.)
I
All,
http://oss.oracle.com/~smushran/.ocfs2-1.2.4-0.1/
This is for users interested in the preliminary patch. The main
issue (not releasing the memory) has been ironed out but there
are some small races we still need to plug. If users want packages
for different kernels/arches, the tarball has
Increase default disk heartbeat timeout. For more refer to the FAQ.
HAWKER, Dan wrote:
Hi All,
Have just installed OCFS2 as a test on a server I am planning on rolling out
a clusterFS onto.
The server(s) will be running FC5 and are standard HP DL360 G5's, have a
Qlogic iSCSI HBA inside (QLA401
Peter Santos wrote:
The backup volume availability requirements shouldn't be so strict
because if the volume is not available, I
would just rather have my RMAN backup fail, and not have the machine
reboot
.. just because of this volume alone.
My question is, is
As per the config, your node names are 'san' and 'mail'.
Are the names the same as the hostname?
Do on both nodes:
# for i in /config/cluster/san/node/*/local ; do LOCAL=`cat $i`; if [
$LOCAL -eq 1 ] ; then echo $i; fi; done;
You should see /config/cluster/san/node/mail/local on mail and
/conf
Currently it supports only one cluster.
Peter Santos wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Folks,
When I installed ocfs2 the first time and setup oracle to work with it,
the clustername defaulted to
"ocfs2". We are testing adding new nodes, but we would like to
never be part of that domain.
Sunil Mushran wrote:
Currently it supports only one cluster.
Peter Santos wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Folks,
When I installed ocfs2 the first time and setup oracle to work
with it, the clustername defaulted to
"ocfs2"
In short, node 1's attempt to connect to node 0 failed. See the
messages file on node 0. If there is no evidence of a connect request,
check if you have some sort of firewall setup. If so, open up the
required port(s).
You could also use tcpdump to trap traffic on the required port /
ethernet int
Refer to the FAQ. The module's kernel version should match
the kernel version.
Lin Shen (lshen) wrote:
Hi,
The kernel I'm using is:
[EMAIL PROTECTED] Desktop]# uname -a
Linux cfs2 2.6.9-42.7.ELsmp #1 SMP Tue Sep 5 18:29:39 EDT 2006 i686 i686
i386 GNU/Linux
So I installed ocfs2-2.6.9-42.ELsm
depmod -a ?
Lin Shen (lshen) wrote:
Switched the kernel to 2.6.9-42.Elsmp, still got the same error.
[EMAIL PROTECTED] Desktop]# uname -a
Linux cfs2 2.6.9-42.ELsmp #1 SMP Wed Jul 12 23:27:17 EDT 2006 i686 i686
i386 GNU/Linux
-Original Message-
From: Sunil Mushran [mailto:[EMAIL
:
I don't see configfs.ko.
[EMAIL PROTECTED] Desktop]# ls /lib/modules/2.6.9-42.ELsmp/kernel/fs/ocfs2/
ocfs2_dlmfs.ko ocfs2_dlm.ko ocfs2.ko ocfs2_nodemanager.ko
-Original Message-
From: Sunil Mushran [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 04, 2007 12:08 PM
To
as TIPC?
o2net code is pretty well contained and isolated. while we have
discussed tipc,
not sure if we ever gave it a serious look.
lin
-Original Message-----
From: Sunil Mushran [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 04, 2007 1:21 PM
To: Lin Shen (lshen)
theoretically yes... but for practical usage go with atleast iscsi
Lin Shen (lshen) wrote:
So w/o shared disk, is it possible to make OCFS2 to work by utilizing
GNBD or etc?
lin
-Original Message-
From: Sunil Mushran [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 04, 2007 2
That and also we've seen similar issues with Broadcom TG3 drivers. We use
Intel E1000 mostly and thus did not experience the same issue.
As far as the configurable net timeouts goes, the patch was added into
mainline on Dec 4th. So it will be available with ocfs2 1.4. We are still
seeing if we ha
Ah yes... you want o2cb service script to be smarter. :)
Edit /etc/init.d/o2cb and remove load_module configfs from the following
line.
It wouldn't make it smarter... just would get it to work.
LOAD_ACTIONS=("load_module configfs"
"mount_fs configfs "'$(configfs_path)'
Lot of ink has been spilled on this subject. ;)
Check out the heartbeat section in the FAQ. One easy solution is to
increase the hb timeout to 60 secs...
O2CB_HEARTBEAT_THRESHOLD = 31
We will leaning towards making that number the default in the 1.4 release.
George Liu wrote:
Both systems cras
You are using two different versions of ocfs2 on the two nodes.
Different enough that they are not network compatible.
It is working as designed.
Consulente3 wrote:
Hi,
I'm new to ocfs2, and in my test's environment, i have:
2 node, becks and vaix
becks can mount ocfs2 fs, but vaix can't.
Whe
Arkadiy Kulev wrote:
Hello ocfs2-users,
I have 2 questions:
1. I forgot the block size and cluster size that I chose
during formatting of the drive. Is there any way I can find it out
afterwards?
# debugfs.ocfs2 -R "stats" /dev/sdX
Look for Cluster Size bits and Block Size bits.
9
Looks to be running out of lowmem.
# date
# cat /proc/meminfo
# cat /proc/slabinfo
Run a script that dumps the above every 1 to 5 mins. That should
help explain the cause.
Brian Sieler wrote:
Using 2-node clustered file system on DELL/EMC SAN/RHEL
2.6.9-34.0.2.ELsmp x86_64.
Config:
O2CB_HEAR
Could be. But I cannot say for sure till I don't get the slab/mem data.
Brian Sieler wrote:
Does this appear to be the same issue as the "OOM Killer" issue
previously reported that would be fixed with ocfs2 1.2.4?
On 1/16/07, Sunil Mushran <[EMAIL PROTECTED]> wrote:
Looks
All,
http://oss.oracle.com/~smushran/.ocfs2-1.2.4-0.2/
The final 1.2.4 should look very close to this drop. We still have one
slippery issue open that we are working on. But, other than that, this
drop is looking good.
The list of patches added post 1.2.4-0.1 is as follows:
r2948: fs - Allow d
1. In SLES10, the /config has been moved to /sys/kernel/config. That's
how it
is on mainline.
2. To monitor heartbeat do:
# watch -d -n2 debugfs.ocfs2 -R "hb" /dev/sdX
This comand will work if you have ocfs2-tools 1.2.2. (Not sure whether
sles10 ships
with 1.2.2 or 1.2.1.) If 1.2.1, do:
# watc
http://oss.oracle.com/projects/ocfs2/dist/documentation/ocfs2_faq.html#NFS
John Lange wrote:
I am attempting to export a 4 node ocfs2 file system via NFS without
much success.
When a client mounts the export file system, /var/log/messages spews
thousands of errors as follows:
kernel: (11514,0)
#define EACCES 13 /* Permission denied */
The messages are harmless. Patch to silence them has already been checked
into the 1.2 repo and mainline git.
Matthew Flusche wrote:
I’m seeing the following errors in my two node cluster. Is this
anything to be concerned with?
Host information:
Red
o2net timeout cannot cause the o2hb panic. The two are totally
different. From the outputs, I would guess o2hb is timing out but
I cannot say for sure till I don't see the full logs.
Andy Phillips wrote:
Its worth pointing out that the o2net idle timer is triggering on the
network heartbeat, whi
rror
messages usually follow. If I'm wrong, please email me directly and help
sort out my understanding.
Andy
On Mon, 2007-01-22 at 10:38 -0800, Sunil Mushran wrote:
o2net timeout cannot cause the o2hb panic. The two are totally
different. From the outputs, I would guess o2hb is timing
This was the lvb issue that was fixed long ago. In the 1.2 tree, it was
fixed in 1.2.2.
2.6.18 should definitely have the fix for this.
davide rossetti wrote:
OS: Fedora Core release 4 (Stentz)
KERNEL: Linux rack1.ape 2.6.17-1.2142_FC4smp #1 SMP Tue Jul 11
22:57:02 EDT 2006 i686 i686 i386 GNU/
ECTED]>
davide rossetti wrote:
On 1/23/07, *Sunil Mushran* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
This was the lvb issue that was fixed long ago. In the 1.2 tree,
it was
fixed in 1.2.2.
2.6.18 should definitely have the fix for this.
it seems it
o2hb is timing out because the io to the device is taking too much time.
Not much one can do other than increase the time out. Say 2mins.
O2CB_HEARTBEAT_THRESHOLD = 61
Consulente3 wrote:
I can reprodute it, every time on heavy IO
I have read this FAQ:
I encounter "Kernel panic - not syncing:
The o2cb script fix is in ocfs2-tools 1.2.2 released Oct 2006.
Ping SUSE for the update.
[EMAIL PROTECTED] wrote:
Using SuSE SP2 Linux running V1.0.8 of OCFS2 and the tools/console
that comes with SP2 distribution.
I am unable to set the* O2CB_HEARTBEAT_THRESHOLD* parameter in the
/etc/sysc
This is not a fs issue. As in the file must be alright. This is a dlm issue.
The fs is asking the dlm to free the lock and the dlm is stuck. How many
nodes do you have? We've fixed a bunch of dlm bugs since what you appear
to be running.
davide rossetti wrote:
I rebooted the two faulty nodes.
no
All,
We are pleased to announce the release of OCFS2 1.2.4-2.
This release addresses the lowmem consumption issue that has plagued
many users.
It also addresses few races in the dlm relating to the lockres migration.
The complete list of changes post 1.2.3 is available here:
http://oss.oracle
It could be that the device name is not the same across the two nodes.
Do:
# mounted.ocfs2 -d
on both nodes. Match the device using the uuid. As in, you
should see a device with the same uuid on both nodes. If not,
then the device is not shared.
If you do see the device on both nodes but with di
The device needs to be shared. As in, both nodes need to be able
to see the same device concurrently.
Refer to iscsi, fiber channel, aoe, etc.
aibolit 66 wrote:
-Original Message-
From: Sunil Mushran <[EMAIL PROTECTED]>
To: aibolit 66 <[EMAIL PROTECTED]>
Date: Mon, 05 Feb 2
m: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Sunil Mushran
Sent: Friday, February 02, 2007 8:29 PM
To: ocfs2-announce@oss.oracle.com; ocfs2-users
Subject: [Ocfs2-users] OCFS2 1.2.4-2 released
All,
We are pleased to announce the release of OCFS2 1.2.4-2.
This release addresses the lowmem
That's the source.
Randy Ramsdell wrote:
Mark Fasheh wrote:
On Tue, Feb 06, 2007 at 10:18:51AM -0500, Randy Ramsdell wrote:
Is source available?
http://oss.oracle.com/projects/ocfs2/dist/files/source/v1.2/ocfs2-1.2.4.tar.gz
--Mark
--
Mark Fasheh
Senior Software
The following patch will address this issue. The fix will be provided
with the next tools release.
Index: libocfs2/include/ocfs2.h
===
--- libocfs2/include/ocfs2.h(revision 1269)
+++ libocfs2/include/ocfs2.h(revision 1270)
@
Means there was a network hiccup that caused Node 1 to fence itself.
The problem is that our default timeout is too low. We have already
addressed this in mainline and are looking to add that patch into 1.2.5.
I am unclear as to your last qs.
Randy Ramsdell wrote:
Hi,
Ok I'll try this again si
The datavolume code is not in mainline. But you should
be able to get Oracle RDBMS to work with it. Ensure the
init.ora paramater filesystemio_options is set to direct_io.
Ivo Maya wrote:
Hi,
I need to mount ocfs2 with datavolume option on open
SuSE 10.2 Machines.
ocfs2 is 1.3.3 version and doe
There are two heartbeats in OCFS2. One on disk and the
other on the network.
Randy Ramsdell wrote:
Sunil Mushran wrote:
Means there was a network hiccup that caused Node 1 to fence itself.
The problem is that our default timeout is too low. We have already
addressed this in mainline and are
What does dmesg say?
Randy Ramsdell wrote:
Hi,
Everything compiled correctly for the ocfs2 package, but so far the
modules will not load with the "well known" module symbol error.
FATAL: Error inserting ocfs2
(/lib/modules/2.6.16.27-0.6-smp/kernel/fs/ocfs2/ocfs2.ko): Unknown
symbol in module,
Appears ocfs2 fs module is not compatible with the other modules
ocfs2_dlm.ko, ocfs2_nodemanager.ko, ocfs2_dlmfs.ko, configfs.ko.
When you build the modules, ensure you copy all of them in /lib/modules/...
before running depmod -a.
Randy Ramsdell wrote:
Sunil Mushran wrote:
What does
No.
Lin Shen (lshen) wrote:
Hi,
Can someone let me know if OCFS2 support the following features.
1. Quota.
2. POSIX ACL
3. Clustered volume manager
Lin
___
Ocfs2-users mailing list
Ocfs2-users@oss.oracle.com
http://oss.oracle.com/mailman/listinfo/o
I do 2.
You could also look into adding the modules to /lib/modules/`uname
-r`/update.
I believe depmod searches that patch before the others.
Randy Ramsdell wrote:
Sunil Mushran wrote:
Appears ocfs2 fs module is not compatible with the other modules
ocfs2_dlm.ko
If a vendor is distributing ocfs2, then the vendor controls the
location. The spec file in our tree is only relevant if we build
the package for that distro.
Randy Ramsdell wrote:
Joel Becker wrote:
On Fri, Feb 09, 2007 at 02:55:37PM -0500, Randy Ramsdell wrote:
I found the issue. T
Are you using 1.2.4? If not, upgrade. We fixed a NFS -ESTALE (-116)
issue in it. Though that the error occurs while mounting is a bit puzzling.
Brandon Lamb wrote:
On 2/9/07, Brandon Lamb <[EMAIL PROTECTED]> wrote:
On 2/9/07, Brandon Lamb <[EMAIL PROTECTED]> wrote:
> Hey
>
> So I installed a ne
It's probably because you are missing some package.
See the FAQ for the list of packages it is dependent on.
Randy Ramsdell wrote:
Hi,
I see that the ocfs2console app for 1.2.2 doesn't have the same menu
items as does the 1.1.0 package. Is the propagate config, check and
repair going to be adde
I meant the solution mentioned by Mark is listed on the ocfs2 home page.
Randy Ramsdell wrote:
I configured a new cluster using 1.2.4-2. This was a custom install by
compiling source.
Any ideas or questions?
The segmentation fault produced with "rm -rf" or any other rm switch :
NOTE: This doe
Yes, this is mentioned on the ocfs2 home page.
Randy Ramsdell wrote:
I configured a new cluster using 1.2.4-2. This was a custom install by
compiling source.
Any ideas or questions?
The segmentation fault produced with "rm -rf" or any other rm switch :
NOTE: This does not happen on a local fi
Use cp. Or use dd with bs=1M.
LOPEZ DIAZ, JORGE wrote:
One server with two hba's, one to old SAN and the other to new one, mounting each fs on
different points and "cp" command with convenient options? May be too slow.
Jorge López Díaz
Técnico de Gestión Inform
Fixed in 1.2.4. SUSE has the patch-fix.
The patch has also been added to mainline.
John Lange wrote:
Yes, the clients are doing lots of creates.
But my question is, if this is a memory leak, why does ocfs2 eat up the
memory as soon as the clients start accessing the filesystem. Within
about 5-1
Do you have the full oops trace?
Nathan Ehresman wrote:
I have a strange OCFS2 problem that has been plaguing me. I have 2
separate OCFS2 clusters, each consisting of 3 machines. One is an
Oracle RAC, the other is used as a shared DocumentRoot for a web
cluster. All 6 machines are in an IBM
That's probably dlm communication. You should be
able to confirm that using ethereal/wireshark.
http://oss.oracle.com/~smushran/.debug/wireshark/
Sebastian Reitenbach wrote:
Hi list,
I just have a quick question. We are experimenting with ocfs2 and linux
heartbeat, using user based heartbeat
OCFS2 has two trees. The 1.2 tree and the git tree.
All new development happens on git head. All bug fixes are typically
worked on the tree that it was detected on. Later, the bug fix is applied
to the other tree. As most of our users are using the 1.2 tree, almost all
bug fixes flow from the 1.2
And you are convinced that drdb's primary-primary is not the
cause for the slowdown. ??
Egon Burgener wrote:
Hi all
We are using a 2 node cluster with drbd 8 (primary/primary state) and
ocfs2. Reading a file on one node while it will be written on the other
node is very slow. Reading a file on
Egon Burgener wrote:
And you are convinced that drdb's primary-primary is not the
cause for the slowdown. ??
Yes, writing a file is fast. Reading a file has no influence on drbd.
We noticed, that reading a big file on one node while the other node
opened that file in RW mode but without wr
Start monitoring /proc/slabinfo and /proc/meminfo. Dump it to a file
every 5-10 mins.
Which version of the rhel4 kernel are you on: uname -a?
Cline, Ernie wrote:
I have a 2 node cluster of HP DL380G4s. These machines are attached via
scsi to an external HP disk enclosure. They run 32bit RH A
t run over the weekend.
-Original Message-
From: Sunil Mushran [mailto:[EMAIL PROTECTED]
Sent: Friday, February 23, 2007 2:25 PM
To: Cline, Ernie
Cc: ocfs2-users@oss.oracle.com
Subject: Re: [Ocfs2-users] OCFS 1.2.4 memory problems still?
Start monitoring /proc/slabinfo and /proc/meminfo.
Hmmm... the last time I saw your numbers, ocfs2's foot print was 15M.
You'll have to do better than that.
Anycase, Luis problem is the relationship between swap and cached buffers.
Which kernel is this?
John Lange wrote:
It seems that ocfs has an unfixed memory leak even in the most recent
vers
Yes, the messages are related. -112 is EHOSTDOWN.
Sebastian Reitenbach wrote:
Hi list,
I am experimenting with ocfs2 (rpm package: 1.2.2-0.2), using linux-ha 2.0.8
(all running on a SLES 10 x86-64, rpm packages from linux-ha.org) for the
heartbeat. The three nodes are connected on a gigabit s
er is 2.6.9-34.ELsmp (RedHat 4.0) as this is
our production cluster and we did not update it recently.
Regards,
Luis
*/Sunil Mushran <[EMAIL PROTECTED]>/* wrote:
Hmmm... the last time I saw your numbers, ocfs2's foot print was 15M.
You'll have to do better than that.
Check out this bug:
http://oss.oracle.com/bugzilla/show_bug.cgi?id=854
José Costa wrote:
Hello,
I'm using 2.6.16.41-SLES10_SP1_BRANCH_20070220135926-smp with OCFS2
1.2.4.
If I start the node1 and then the node2... everything works. If I
reboot the node1, it gives this error to node2 and I ca
Alexei_Roudnev wrote:
(Btw, some Oracle engineers are saying that they are going to drop OCFSv2
certification.)
I would rather you not use this channel to spread FUD.
___
Ocfs2-users mailing list
Ocfs2-users@oss.oracle.com
http://oss.oracle.com/ma
Means that the volume is in use on at least one node in the cluster.
If you were using the native o2cb heartbeat, the following command
would have shown the heartbeating node.
# watch -d -n2 "debugfs.ocfs2 -R \"hb\" /dev/sdX"
Ping SUSE to find out the details when using ocfs2 with linux-ha.
Seb
This specific issue was long addressed in ocfs2 1.2.2. The fix is available
on sles9 with 2.6.5-7.283.
Alexei_Roudnev wrote:
Saw it few times until I unmounted FS on all nodes, run fsck (show nothing)
and then mounted back:
Do we have any errors/bugs, explaining this:
Mar 1 06:26:42 spproddoc0
If you want to use the mainline version of ocfs2, use a raw partition
for the voting disk and ocr.
For the db, specify filesystemio_options=directio in init.ora.
nirmal tom wrote:
hi,
i am trying to install oracle rac on fedora core 6 through iscsi.
when i try to mount,
mount.ocfs2: Invalid ar
All,
We are pleased to announce the release of OCFS2 Tools 1.2.3. This release
is fully compatible with the OCFS2 1.2.x and the OCFS2 bundled with the
mainline Linux kernel 2.6.20 (and earlier).
The summary of changes in this release are as follows:
* Backup super block support added
* Local mou
How many nodes in the cluster?
Marcus Alves Grando wrote:
Hi list,
I have some problems testing ocfs2. My test consist in:
#server1: dd if=/dev/random of=/ocfs2_1/test &
#server1: dd if=/dev/random of=/ocfs2_2/test &
#server1: dd if=/dev/random of=/ocfs2_3/test &
...
#server1: dd if=/dev/rando
N-COLUMN
Stop the tcpdumps and make them available to me via some ftp site or
whatever.
Also, file a bugzilla for tracking purposes.
Marcus Alves Grando wrote:
Sunil Mushran wrote:
How many nodes in the cluster?
Four.
Marcus Alves Grando wrote:
Hi list,
I have some problems testing ocfs2.
Well, kswapd is supposed to flush the caches. As in, the vm
controls the lifetime of the inodes in the inode_cache not ocfs2.
All ocfs2 can do is free the memory associated with the inode when
asked to. And it does that when you manually flush the cache. Qs is
why the vm is not doing it on its ow
x27;t have much hopes that it works., More likely it
will work well since SLES10 SP3, as usual (all, SLES9 and SLES8, became 100%
reliable starting with SP3 approximately).
- Original Message -
From: "John Lange" <[EMAIL PROTECTED]>
To: "Sunil Mushran" <[
If you are running a prod shop, you should looking into buying support.
John Lange wrote:
On Mon, 2007-03-05 at 13:46 -0800, Sunil Mushran wrote:
Well, kswapd is supposed to flush the caches. As in, the vm
controls the lifetime of the inodes in the inode_cache not ocfs2.
All ocfs2 can do
hit harsh and disrespectful to say the least. Which is never really
appreciated. A little bit of respect and more constructive feedback usually
goes a very long way. Everyone is tryong their best.
-Original Message-
From: "Alexei_Roudnev" <[EMAIL PROTECTED]>
To: &q
File a bugzilla with the messages from all three nodes. Appears
node 2 went down but kept heartbeating. Strange. The messages
from node 2 may shed more light.
Andy Kipp wrote:
We are running OCFS2 on SLES9 machines using a FC SAN. Without warning both nodes will become unresponsive. Can not acces
The config error I would imagine would be that you defined two
different clusters, each not having the other node, and that the two
nodes have the same node number in both clusters. If so, the disk hb
would have detected this error. It would have spewed error messages
indicating that "some other n
Have you tried to do alt-sysrq-t on the "dead" node? The stack traces
will be
of great help.
Also, even though this could be the same as #819, I would still recommend
filing a new bug with all the messages files. Even though that will take
some
of your time, it will be much easier to keep track
bversion from development holds support for it?Any other
ways of using ocfs2?
thanks for the response
regards,
Nirmal Tom.
From: Sunil Mushran <[EMAIL PROTECTED]>
To: nirmal tom <[EMAIL PROTECTED]>
CC: ocfs2-users@oss.oracle.com
Subject: Re: [Ocfs2-users] data volume option,
ginal Message -
From: "Sunil Mushran" <[EMAIL PROTECTED]>
To: "nirmal tom" <[EMAIL PROTECTED]>
Cc:
Sent: Tuesday, March 13, 2007 9:28 AM
Subject: Re: [Ocfs2-users] data volume option, is it present in current
versionof ocfs2
Just specify block devices for vot
Yes. 1.2.5 will have the configurable network timeout.
Randy Ramsdell wrote:
Hi,
We are planning an upgrade of an ocfs2 cluster and I wanted to clarify
something first. Is 1.2.5 going to include a variable to set the network
timeout? This seems to be important as we have had to move processes
mand for memory, why shouldn't caches stay in
memory? That's the entire point of caches. If we start OOM killing
processes due to the caches taking all the memory, that's absolutely a bug.
Here is what Sunil Mushran from Oracle had to say about the issue:
Well, kswapd
Peter Santos wrote:
"Mar 16 13:38:02 dbo3 kernel: (3712,3):o2hb_do_disk_heartbeat:963 ERROR: Device
"sdb1": another node is
heartbeating in our slot!"
Usually there are a number of other errors, but this one was it.
If this was one isolated er
e.
Stephan Hendl wrote:
fine, we appreciate this variable as well. Will there be a procedure for
a rolling upgrade? We are using 1.2.4 in a production environment. From
1.2.3 to 1.2.4 there whole cluster had to be offline during upgrade...
;-(
Stephan
Sunil Mushran <[EMAIL PROTECTED]> sc
The patch fix for this missed the 2.6.20 window.
The following link has all the relevant patches atop 2.6.20.
http://git.kernel.org/?p=linux/kernel/git/mfasheh/ocfs2.git;a=log;h=2.6.20_fixes
Apply all in order starting from one after the official 2.6.20.
Incidentally the fix you require is the
The kernel has changed much. 1.2 will not build against 2.6.20.
We have updated autogen to handle rhel5, but that's still 2.6.18.
What are you trying to achieve? Why not use the ocfs2 modules
shipped natively with fc6? If you want to run database, specify
filesystemio_options=odirect in init.ora.
Please file a bug in oss.oracle.com/bugzilla with the full strack trace.
The kernel BUG line looks wrong or at least incomplete. Is this the
full trace? If you have hit this on more than one occasion please do mention
that in the bugzilla. Maybe even upload the stack trace from those
instances t
#define ECONNRESET 104 /* Connection reset by peer */
#define EPIPE 32 /* Broken pipe */
Harmless. But do file a bug. oss.oracle.com/bugzilla. We should not be
printing the ERROR. It should be handled by the userspace.
Stephan Hendl wrote:
Hi,
I'm using a 4 node cluster
You'll run into the size-256 slab explosion on sles9 sp3.
That issue was addressed in 1.2.4. sp3 ships 1.2.3.
Alexei_Roudnev wrote:
OCFSv2 @ SLES9 Sp3 build 283 is relatively stable. I am running your test on
2 hosts now (create files from 2 hosts, and delete them with some delay from
host1 by
How long does it take for the node to die?
File a new bugzilla with the following info.
date >> /tmp/info.txt
iostat -x 1 3 >> /tmp/info.txt
vmstat 1 3 >> /tmp/info.txt
top -b -n 1 | head -50 >> /tmp/info.txt
ps -elf>> /tmp/info.txt
cat /pro
ocfs2 init script is mounting devices listed in /etc/fstab.
Check the device names. If you are mounting by device name,
the name may have changed. If so, fix the device name and also
look up mount by label in the docs.
John E wrote:
Hi All,
I needed to rebuild the operating system on one of the
In ocfs2, the default network timeouts are too low. The patch fix
to make this timeout configurable is available in sles10 sp1.
David Miller wrote:
Good afternoon all;
I'm planning on implementing a shared storage solution for a primary
and backup oracle server in the near future.
We can't a
The packages have been spun and have been in testing since last week.
We should get the green light sometime later this week.
Randy Ramsdell wrote:
Hi,
We were having issues with an ocfs2 cluster and I was really thinking
the instability was related to ocfs2. In retrospect, it was an ocfs2
prob
This is a known issue on SLES10. Ping Novell for the update.
Eli Criffield wrote:
Whenever i mount my shared ocfs2 volume on the second node, the
primary kernel panics.
I have SLES10 xen guests both able to access the same /dev/sdc1.
My /etc/ocfs2/cluster.conf
cluster:
node_cou
You may want all nodes to be on the same kernel version. 2.6.5-7.283.
But that's not the problem. There was a bug in tunefs.ocfs2 1.1.4. The
jbd superblock was not endian correct. Was fixed shortly thereafter.
You can fix this issue as follows:
1. Use yast to upgrade to ocfs2-tools 1.2.1 (or hig
ocfs2 currently lets vfs handle fcntl locking.
Jeff Fookson wrote:
I am currently testing ocfs2 for use in a two-node cluster that will
run the Cyrus imapd and am having issues
that seem to be related to occasionally long times being needed while
the software blocks waiting to get a writelock
v
You will have to provide more information. If you
have a netconsole server configured, it would have the details.
Else, I would recommend you configure one to catch the
messages during fence. We have to see the deduce for the fence
to determine the actual problem.
enohi ibekwe wrote:
Is this als
All,
We are pleased to announce the release of OCFS2 1.2.5 and
OCFS2 Tools 1.2.4.
This release provides the long awaited configurable network
timeout feature. This feature was originally written by Andrew
Beekhof and ported to the OCFS2 1.2 tree by Jeff Mahoney, both
of Novell/SUSE.
The complet
mkdir /u01 or /u02
As in, it appears you are missing the mount directory.
Zosen Wang wrote:
I try to install 2 nodes RAC in Linux 2.6.9.-22.EL by using Jeffery
Hunter’s paper. I am getting problem to mount the ocfs2 file system.
The following is mount command error output:
[EMAIL PROTECTED] ~]
201 - 300 of 1424 matches
Mail list logo