Comments for nxnt.org http://www.nxnt.org New eXperiences, New Technologies Wed, 09 Nov 2011 15:19:10 +0000 hourly 1 http://wordpress.org/?v=3.3 Comment on Multipath linux and EMC CLARiiON by MultiPath configuration on RHEL5 and Clariion CX-300 | web technical support http://www.nxnt.org/2010/08/multipath-linux-and-emc-clariion/#comment-156 MultiPath configuration on RHEL5 and Clariion CX-300 | web technical support Wed, 09 Nov 2011 15:19:10 +0000 http://www.nxnt.org/?p=8#comment-156 [...] followed instructions from this site (editing /etc/multipath.conf), but i failed after multipath -ll – the output was [...] [...] followed instructions from this site (editing /etc/multipath.conf), but i failed after multipath -ll – the output was [...]

]]> Comment on Multipath linux and EMC CLARiiON by ukh http://www.nxnt.org/2010/08/multipath-linux-and-emc-clariion/#comment-132 ukh Wed, 02 Nov 2011 10:13:49 +0000 http://www.nxnt.org/?p=8#comment-132 On CentOS/RHEL 6, try replacing prio_callout "/sbin/mpath_prio_emc /dev/%n" with prio emc On CentOS/RHEL 6, try replacing

prio_callout “/sbin/mpath_prio_emc /dev/%n”

with

prio emc

]]> Comment on RedHat Cluster howto by ztank http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-107 ztank Thu, 20 Oct 2011 09:22:17 +0000 http://www.nxnt.org/?p=21#comment-107 That's a good news! thanks. So there is no real error but only the volume tag mechanism working fine. I also took a look at the <code>vgchage</code> man page and apparently using <code>/sbin/lvm.static vgchange -a ey</code> or <code>/sbin/lvm.static vgchange -c y</code> can be a viable solution, but I am not convinced. The <code>vgchange</code> man says: <code> -a, --available [e|l]{y|n} Controls the availability of the logical volumes in the volume group for input/output. In other words, makes the logical vol- umes known/unknown to the kernel. If clustered locking is enabled, add âeâ to activate/deactivate exclusively on one node or âlâ to activate/deactivate only on the local node. Logical volumes with single-host snapshots are always activated exclusively because they can only be used on one node at once. -c, --clustered {y|n} If clustered locking is enabled, this indicates whether this Volume Group is shared with other nodes in the cluster or whether it contains only local disks that are not visible on the other nodes. If the cluster infrastructure is unavailable on a particular node at a particular time, you may still be able to use Volume Groups that are not marked as clustered. </code> What do you think? <strong>PS:</strong> Why don't you add a Preview button (if available) before actually posting the comment. It might be of some help especially if people uses HTMLformatting tags like me :) That’s a good news! thanks.
So there is no real error but only the volume tag mechanism working fine. I also took a look at the vgchage man page and apparently using /sbin/lvm.static vgchange -a ey or /sbin/lvm.static vgchange -c y can be a viable solution, but I am not convinced.

The vgchange man says:


-a, --available [e|l]{y|n}
Controls the availability of the logical volumes in the volume
group for input/output. In other words, makes the logical vol-
umes known/unknown to the kernel.

If clustered locking is enabled, add âeâ to activate/deactivate
exclusively on one node or âlâ to activate/deactivate only on
the local node. Logical volumes with single-host snapshots are
always activated exclusively because they can only be used on
one node at once.

-c, --clustered {y|n}
If clustered locking is enabled, this indicates whether this
Volume Group is shared with other nodes in the cluster or
whether it contains only local disks that are not visible on
the other nodes. If the cluster infrastructure is unavailable
on a particular node at a particular time, you may still be
able to use Volume Groups that are not marked as clustered.

What do you think?

PS: Why don’t you add a Preview button (if available) before actually posting the comment. It might be of some help especially if people uses HTMLformatting tags like me :)

]]> Comment on RedHat Cluster howto by nicola http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-106 nicola Wed, 19 Oct 2011 09:55:20 +0000 http://www.nxnt.org/?p=21#comment-106 Hi, I was looking inside <code>/etc/init.d/netfs</code> script and I extract these lines: <code>if [ -x /sbin/lvm.static ]; then if /sbin/lvm.static vgscan > /dev/null 2>&1 ; then action $"Setting up Logical Volume Management:" /sbin/lvm.static vgchange -a y fi fi</code> if you run <code>vgchange -a y</code> you will always get errors because of volumes tags. So you can ignore such error, or you can comment the line in the script. Hi,
I was looking inside /etc/init.d/netfs script and I extract these lines:

if [ -x /sbin/lvm.static ]; then
if /sbin/lvm.static vgscan > /dev/null 2>&1 ; then
action $"Setting up Logical Volume Management:" /sbin/lvm.static vgchange -a y
fi
fi

if you run vgchange -a y you will always get errors because of volumes tags.

So you can ignore such error, or you can comment the line in the script.

]]> Comment on RedHat Cluster howto by ztank http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-105 ztank Tue, 18 Oct 2011 14:47:33 +0000 http://www.nxnt.org/?p=21#comment-105 Very nicely done article, thanks it helped me a lot! I have a question about the <code>volume_list</code> parameter in <code>/etc/lvm/lvm.conf</code>. In my cluster I have no vg00, in fact my clustered installation (at the moment) does not use lvm to handle OS partition. On the other hand I do use lvm to handle quorum disk and shared application disk, hence my <code>volume_list</code> line is something like this: <code>volume_list = [ "vg_qdisk", "@node-a" ]</code> This works like a charm and thanks to the tagging mechanism (<code>@</code>) the shared disk is only loaded on the active node avoiding data corruption. The mechanism is such great that even <code>/dev/vg.../...</code> links to devices are removed from the passive node! My only concern is about the boot process... when starting the node-a for example I get an error message while the lvm service is starting probably because the tagged volume (@node-a) cannot be handled unless the cluster decide for it to be handled. Is this assumption correct? Should I move the lvm service later in the boot process to fix this? The error reported during the boot process is: <code>Setting up Logical Volume Managment ... [FAILED]</code> Any thoughts? Very nicely done article, thanks it helped me a lot!

I have a question about the volume_list parameter in /etc/lvm/lvm.conf. In my cluster I have no vg00, in fact my clustered installation (at the moment) does not use lvm to handle OS partition. On the other hand I do use lvm to handle quorum disk and shared application disk, hence my volume_list line is something like this:

volume_list = [ "vg_qdisk", "@node-a" ]

This works like a charm and thanks to the tagging mechanism (@) the shared disk is only loaded on the active node avoiding data corruption. The mechanism is such great that even /dev/vg.../... links to devices are removed from the passive node!
My only concern is about the boot process… when starting the node-a for example I get an error message while the lvm service is starting probably because the tagged volume (@node-a) cannot be handled unless the cluster decide for it to be handled. Is this assumption correct? Should I move the lvm service later in the boot process to fix this?
The error reported during the boot process is:

Setting up Logical Volume Managment ... [FAILED]

Any thoughts?

]]> Comment on RedHat Cluster howto by Dawood M. G. http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-77 Dawood M. G. Sat, 17 Sep 2011 18:27:34 +0000 http://www.nxnt.org/?p=21#comment-77 Hi Nicola, just for registering in the history ;-) and after tries and tries it seems that MSA60/70 is not capable of doing clustering under any OS except HPUX! In fact it does not support parallel writes from both nodes simultaneously! My problem was the point that when both of the nodes wanted to write at the same time on the same LUN on MSA60 (i.e. for the quorum disk), the system became unstable (because of the above problem), I don't know why hp does not add this capability to its drivers on Linux, maybe just to force users to buy much more expensive MSA2xxx series. It was all I found, I would be glad if someone can complete these findings in future. regards, Dawood Hi Nicola,
just for registering in the history ;-) and after tries and tries it seems that MSA60/70 is not capable of doing clustering under any OS except HPUX! In fact it does not support parallel writes from both nodes simultaneously! My problem was the point that when both of the nodes wanted to write at the same time on the same LUN on MSA60 (i.e. for the quorum disk), the system became unstable (because of the above problem), I don’t know why hp does not add this capability to its drivers on Linux, maybe just to force users to buy much more expensive MSA2xxx series.
It was all I found, I would be glad if someone can complete these findings in future.

regards,
Dawood

]]> Comment on RedHat Cluster howto by nicola http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-64 nicola Tue, 30 Aug 2011 07:21:13 +0000 http://www.nxnt.org/?p=21#comment-64 Uhm, this is strange. Could you send me a link of your cluster.conf file? (please use pastebin) Uhm, this is strange. Could you send me a link of your cluster.conf file? (please use pastebin)

]]> Comment on RedHat Cluster howto by Dawood M. G. http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-60 Dawood M. G. Fri, 19 Aug 2011 23:29:42 +0000 http://www.nxnt.org/?p=21#comment-60 Thanks Nicola for your answer, these hours I have had a light on the whole story of clustering in RHEL, just one problem remained after all the config done: I have the MSA60 attached to two P812 HBAs on the servers, and the file system of the array is ext3, however when I 1- update a file on one node (make some changes on it), 2- reallocate the storage to node 2, 3- mount the FS on node 2, the changes made in the file are not shown. I googled a little and found that I can use for example 'sync' & 'echo 3 > /proc/sys/vm/drop_cache'. Now my question is: Is there any other way to overcome this problem on ext3? something less aggressive? thanks in advance for your replies... Thanks Nicola for your answer, these hours I have had a light on the whole story of clustering in RHEL, just one problem remained after all the config done:
I have the MSA60 attached to two P812 HBAs on the servers, and the file system of the array is ext3, however when I 1- update a file on one node (make some changes on it), 2- reallocate the storage to node 2, 3- mount the FS on node 2, the changes made in the file are not shown.
I googled a little and found that I can use for example ‘sync’ & ‘echo 3 > /proc/sys/vm/drop_cache’. Now my question is: Is there any other way to overcome this problem on ext3? something less aggressive?
thanks in advance for your replies…

]]> Comment on RedHat Cluster howto by nicola http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-59 nicola Fri, 19 Aug 2011 15:50:23 +0000 http://www.nxnt.org/?p=21#comment-59 1. yes, using locking_type=1 is correct here, because we don't use clvmd. 2. This is the expected behaviour. The logical volumes should be active only on one node a time (this avoid concurrent access to the LVs that may corrupt datas) 3. Services may contain different resources. In this case the service contains an IP address and some filesystems. 4. Try to crash the node where the service is running and see if the cluster behave as expected. (to crash a node type as root: "echo c > /proc/sysrq-trigger") Nicola 1. yes, using locking_type=1 is correct here, because we don’t use clvmd.
2. This is the expected behaviour. The logical volumes should be active only on one node a time (this avoid concurrent access to the LVs that may corrupt datas)
3. Services may contain different resources. In this case the service contains an IP address and some filesystems.
4. Try to crash the node where the service is running and see if the cluster behave as expected. (to crash a node type as root: “echo c > /proc/sysrq-trigger”)

Nicola

]]> Comment on RedHat Cluster howto by Dawood M. G. http://www.nxnt.org/2010/09/redhat-cluster-howto/#comment-58 Dawood M. G. Fri, 19 Aug 2011 15:21:53 +0000 http://www.nxnt.org/?p=21#comment-58 I did the installation based on your guid with 2 DL380 NAS Heads and 1xMSA60 Dual Domain-ed and CentOS 5.6 with 2.6.18-194.el5 kernel (mpath for hp just works on kernels of 5.5 or less), there are some questions, I would be glad if you kindly answer me: 1- In all guides they use "locking_type=3" -I think they use clvmd based config-, is it correct to use locking_type=1 here? 2- After setting up everything my HAd logical volume is active on one node and inactive on the other node, is it correct? when I activate it on the node 2 it becomes inactive on the 1st node. (however the qdisk is active on both nodes) 3- there is an IP mentioned in the service part of cluster.conf, which IP should I choose for it? a new IP in the subnet of heartbeat interfaces? 4- How can I check if the configuration is correct? What is the test for this configuration? Regards, Dawood I did the installation based on your guid with 2 DL380 NAS Heads and 1xMSA60 Dual Domain-ed and CentOS 5.6 with 2.6.18-194.el5 kernel (mpath for hp just works on kernels of 5.5 or less), there are some questions, I would be glad if you kindly answer me:
1- In all guides they use “locking_type=3″ -I think they use clvmd based config-, is it correct to use locking_type=1 here?
2- After setting up everything my HAd logical volume is active on one node and inactive on the other node, is it correct? when I activate it on the node 2 it becomes inactive on the 1st node. (however the qdisk is active on both nodes)
3- there is an IP mentioned in the service part of cluster.conf, which IP should I choose for it? a new IP in the subnet of heartbeat interfaces?
4- How can I check if the configuration is correct? What is the test for this configuration?

Regards,
Dawood

]]>