Silly error reported by cmcheckconf

[Ralph Grothe]

Hello SG experts,

I wish to set up a new cluster.

After having done the prerequisites (i.e. configuring the nodes, truted hosts, shared VGs, and editing and distribution of cmquerycl generated cmclconf.ascii) I tried to run a check from the configuratin and distribution node.
Here is the bewildering error message I get:

# cmcheckconf -v -C /etc/cmcluster/cmclconf.ascii

Checking cluster file: /etc/cmcluster/cmclconf.ascii
Checking nodes ... Done
Checking existing configuration ...
Done
Gathering configuration information ............ Done

Warning: The disk at /dev/dsk/c7t2d0 on node europa does not have an ID.
Warning: The disk at /dev/dsk/c7t2d0 on node ganymed does not have an ID.
Warning: Disks which do not have IDs cannot be included in the topology descript
ion.
Use pvcreate(1m) to give a disk an ID.
Error: Device is not found on any nodes in the cluster.
cmcheckconf : Unable to reconcile configuration file /etc/cmcluster/cmclconf.as
cii
with discovered configuration information.


What causes me some confusion is that the mentioned devices are but the CDROM driver interfaces on both nodes:

# ioscan -knfCdisk|tail -2
disk 32 10/12/5.2.0 sdisk CLAIMED DEVICE TOSHIBA CD-ROM XM-5401
TA
/dev/dsk/c7t2d0 /dev/rdsk/c7t2d0

# remsh europa /usr/sbin/ioscan -knfCdisk\|tail -2
disk 32 10/12/5.2.0 sdisk CLAIMED DEVICE TOSHIBA CD-ROM XM-5401
TA
/dev/dsk/c7t2d0 /dev/rdsk/c7t2d0

[Sanjay_6]

Hi Ralph,

this is quite normal and can be ignored. I also get the same error when i do a cmquerycl / cmcheckconf.

Hope this helps.

Regds

[Helen French]

Hi,

Check this out:

http://us-support.external.hp.com/cki/bin/doc.pl/sid=ca5644460681a0880e/screen=ckiDisplayDocument?docId=200000054861742


HTH,
Shiju

[Uday_S_Ankolekar]

Hi,
This is common symptom that is NOT a bug. It just informs that you have discs that have not had pvcreate run on them, and typically these will be CDROM drives, DVD's etc, but will also include any disc that you have not done pvcreate on.

Check this pacth to address this PHSS_24033

Goodluck
-USA..

[John Palmer]

It's only a warning that you can safely ignore.

cmcheckconf and cmapplyconf check that all disks have been pvcreated - including the CD/DVD drives unfortunately.

Regards,
John

[Magdi KAMAL]

Hi Ralph,

Your command is :

# cmcheckconf -v -C /etc/cmcluster/cmclconf.ascii

But I would write, up to the syntaxe of cmcheckconf :

# cd /etc/cmcluster
# cmcheckconf -v -C /etc/cmcluster/cmclconf.ascii -P /_01.conf -P /_01.conf .... -P

/_01.conf

depends how much packages you have !


HTH.

Magdi

[melvyn burnard]

Ok, the message about the disk id's is just what it says, a warning. There is no problem with this.
What seems to have been missed in your discussion is this line:

Error: Device is not found on any nodes in the cluster.


This is what is the actual problem.
I will hazard a guess here and say that you have installed ServiceGuard 11.08, possibly 11.09, and that you have ALL disks in the cluster (including root discs) shared.
Am I correct?

If so you definitely need to patch Sg as this was a known issue.

[Sanjay_6]

Hi Melvyn,

the device referred to above is the internal CDrom drive and is not available on the shared bus. Hence the same cannot be seen from both the systems.

It does not look like a problem to me that needs a patch.

Hope this helps.

Regds

[melvyn burnard]

You are incorrect! read the output more closely.

>Warning: The disk at /dev/dsk/c7t2d0 on node europa does not have an ID.
>Warning: The disk at /dev/dsk/c7t2d0 on node ganymed does not have an ID.
>Warning: Disks which do not have IDs cannot be included in the topology descript
ion.
>Use pvcreate(1m) to give a disk an ID.


The above are the warnings and are to be expected.


>Error: Device is not found on any nodes in the cluster.

This is NOT a warning, but an ERROR! and is the actual problem I believe he is seeing.

[Sanjay_6]

Hi Melvyn,

I wasn't able to locate the patch to fix this problem. Do you have the patch name. I did a search but it didn't come up with anything.

Thanks

[Ralph Grothe]

Hi all,

thank you very much for your replies.

Actually it weren't the warnings about the CDROM that prevented me from generating a clusterconf binary, but rather the error as Melvyn correctly realized.

Thanks Melvyn for your hint.
This seems really to point to the right direction.
You are absolutely correct in your assumption about *all* disks being shared.
That I also didn't realize when I was "given" the two boxes with the order to install OS, DBMS, SG etc on them.
Thus I almost inadvertently erased my Root VG while attempting to mirror on the second node :-}
I think I will have to ask HP Support about the needed patch to overcome the SG configuration error.

[melvyn burnard · Accepted solution]

If it is SG 11.08, you will need PHSS_19508, or if it is SG 11.09, then you need the very latest patch, PHSS_25499.
It will be one of these two, as this problem manifested itself in 11.08, and was also in 11.09

These are both available from the patch download area of the ITRC, but they are large (in the order of 21Mb I think).

[Ralph Grothe]

Hi Melvyn,

many, many thanks for supplying me with the needed patch.
You were indeed much quicker with the remedy than the HP Support call back I've been waiting for :-)

Having installed the missing patch fixed my problem immediately, and I could successfully finish my cluster configuration.

Sorry for my belated assignment of points, but yesterday the HP webserver continously kicked me out when trying to click the assign button with the dreaded "could not login" screen :-(
This is really a nuisance the webserver's maintainers should do something about.
Luckily I could be logged in this morning while, I guess, the Americas are still asleep :-)

Regards
Ralph