topic Re: Cluster_Config question in Operating System - OpenVMS

Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 11:18:32 GMT

Hello all,
currently I have 2 node cluster DS25 (Alpha workstation containing system disk) and XP 1000 (Satellite node) connected via Ethernet Lan

Now I am adding EV68 (Alpha Workstation)

The current cluster config procedure is asking the following questions (two)

Question1:
Will a local disk on SIZZLE be used for paging and swapping (Y/N)? y
Does SIZZLE have any RFxx disks [N]? ?

Answering YES to this question means that SIZZLE will
temporarily serve its RFxx disks to the rest of the cluster
so that this procedure can create local page and swap files.

I am not sure what is RFxx disks
Can anyone please let me know ?

Second Question:

The following workstation windowing options are available:

1. No workstation software
2. DECwindows Workstation Software

Enter choice [1]:

Now current clustered system disk node DS25 is a workstation running Dec windows so I said Yes. But any suggestions comments will be helpful.

Thanks in advance.

Re: Cluster_Config question

Uwe Zessin — Thu, 31 Mar 2005 11:33:31 GMT

RFxx disks used a "DSSI" (Digital Storage Systems Interconnect) interface. It was modelled after the early SCSI implementations, but ran a different protocol (MSCP). It is highly unlikely that you have these disks (look for a KFPSA adapter ;-).

It is easier to install the DECwindows software during system setup, so I'd say you did it the right way. You can always disable it later on should you decide for some reason.

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 12:15:18 GMT

Thanks for the reply UWE,

Now I went through the whole clustering process and the satellite node got rebooted by the current clustered node

However during the rebooting of the Satellite node

It gives 3 messages (repeatedly)

Message from user SYSTEM on XXXXX
Event: Too Few Servers Detected from:Node LOCAL:.XXXXX DTSS,

at: 2005-03-31-12:09:21.444-05:00Iinf
Number Detected = 0
Number Required = 1
eventid XXXXXXX....
entityid XXXXXx...
streamUid xxxxx...

This message is for XXXXX = XP1000,DS25 (currently containing system disk) and EV68 (new node)

After cluster_config was run it said configuration done successfully.

I am not sure what I need to do now?

Do I need to update anything in XP1000 (satellite node) or DS25 (currently containing system disk)

Re: Cluster_Config question

Karl Rohwedder — Thu, 31 Mar 2005 12:29:14 GMT

This comes from DECnet/PLUS. DTSS is the Distributed Time Services (ala NTP in the TCPIP world) and it moans about not being able to find a time source.
You can disable DTSS with some NCL magic, sorry but I have no VMS at my hands in the moment and my memory has this leak...

mfg Kalle

Re: Cluster_Config question

Uwe Zessin — Thu, 31 Mar 2005 12:41:04 GMT

That leak can easily plugged by a search for DTSS ;-)

One useful link:
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=694612

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 13:59:51 GMT

Hello Karl and Uwe,
First of all thanks a lot for getting back to me so soon. The posting gave me the idea that things are not that wrong with my first cluster setup :).

Okay regarding disabling DTSS, I had few issues and concerns.I am stating below exactly how things are going.

Currenltly the DTSS message kept on comming so once I read your message I gave CTrl /C and suddenly it went to the "textual blue login screen" and when I logged in it gave

%LICENSE-I-NOLICENSE,no license is active for this software product
%
Now should I do the process of diabling DTSS as Uwe has written down on the "new terminal" or on the clustered node( containing system disk)? The reason I am asking is because I don't understand what will happen to the current Satellite node (XP 1000). I mean during boot process XP 1000 never gives such message and it directly goes to normal login screen and not the "textual blue screen".

I apologize for asking too many detailed question.

Thanks again in advance

Re: Cluster_Config question

Uwe Zessin — Thu, 31 Mar 2005 14:08:59 GMT

It sounds like you did not enter the licenses (VMS, VMSCLUSTER, DECNET, DW-MOTIF, ?) for the satellite node, yet. You can login to your 'server' and run
$ @sys$update:vmslicense
Then, just reboot your satellite.

I am surprised that a 'CONTROL-C' brought you to the login screen. Do you recall if there was a dialog on the screen before those DTSS messages forced it to scroll off?

Re: Cluster_Config question

Uwe Zessin — Thu, 31 Mar 2005 14:13:02 GMT

Oh, by the way...

there is no need to apologize. These forums are about helping each other. If you provide us with many details, it will make it easier for us to help you.

At least for me I can say that even if I have been doing this for many years I very often learn something new or at least can refresh my memory ;-)

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 14:25:15 GMT

Hi Uwe,
*******
Regarding license, I had installed all the PAks except two
1) OVMS/A smp Asver1200/DS 20 (Product Name :Openvms Alpha)
2) OVMS Alpha Base DS20/DS20E/DS25 LIC (Product Name:OPENVMS-ALPHA

because they had Key Options: NO_SHARE, so I had directly registered them on the "new EV68" when I had booted it independently. Maybe I need to register them. Sys update was the first thing I tried to do but did not make sense in terms of units required and assigned.

******
Regarding CTrl/C message, Unfortunately I did not notice the last message before that.

So right now I just have the "blue terminal screen" and as explained by the forum links...Opcom messages of DTSS, keep coming

Let me know what you think

Re: Cluster_Config question

Ian Miller. — Thu, 31 Mar 2005 14:54:24 GMT

You register paks in the licence database on the cluster shared system disk.
SYS$COMMON:[SYSEXE]LMF$LICENCE.LDB
(you are booting from a common disk in this configuratiion?). Use
SYS$UPDATE:VMSLICENSE to register paks - you can do this on one of the other nodes or on the new node.

Disable DTSS on all nodes by
COPY SYS$COMMON:[SYSMGR]NET$LOGICALS.TEMPLATE SYS$COMMON:[SYSMGR]NET$LOGICALS.COM
Then edit to
$ DEFINE/SYSTEM NET$DISABLE_DTSS 1

this will disable dtss on next boot.
to disable on current system
$ MCR NCL DISABLE DTSS

Re: Cluster_Config question

Uwe Zessin — Thu, 31 Mar 2005 14:58:59 GMT

The OPENVMS-ALPHA licenses are the base licenses for the operating system and you need them. When you enter the data, you get asked for the nodename this license is to be assigned too. NO_SHARE is a hint for the license management that such a license must not be loaded on any other node.

If you reboot your server or enter a
$ LICENSE RELOAD

you will see on the next
$ SHOW LICENSE

that most licenses like DVNETEND or DW-MOTIF have been combined. Don't worry - that is OK.

I would reboot the satellite to see at which state the system is stopping. You can try to do that from your server:

$ run sys$system:sysman
SYSMAN> set environment /node=NAME-OF-SATELLITE
SYSMAN>

I can't test it right now, but I think the next command is:
SYSMAN> shutdown node /automatic_reboot

If that doesn't work, I am afraid you have to manually reset your satellite or do a power-cycle.

Good luck and/or have fun! I'm going to bed now...

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 15:50:13 GMT

Hello Uwe and Ian,
Thanks for answering my questions especially in precise detail. So far I have installed licenses and i disabled DTSS using
$mcr ncl disable dtss
option so that it is only disabled in the newly clustered node

After doing the disabling I got the followin message on the "blue screen"

Node 0 DTSS
%........OPCOM ........%
Message from user SYSTEM on SIZZLE (new node)
EVENT: DISABLE Completed from:
Node Local:.SIZZLE DTSS,
AT : 2005-03-31-15:23:...
eventid...
entityid...
....

Then back to
$ prompt

however after that I did get I think 3 messages exactly same DTSS as I had posted before but now only from the DS25 and XP1000 nodes

I am now going to reboot SIZZLE (only) using Uwe's suggested commands.

Ian you mentioned about changing it in the whole cluster (common system disk). Would that create any problem in currently working setup. I am just concerned that I might bring down the whole cluster (As you had mentioned in the http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=847365) that we need some synchronisation.

Is there a way to check how XP1000 is setup and exactly simulate that since it always boots up automatically and without any DTSS messages

Please let me know what you think. Thanks once again for the details.

Nipun

Re: Cluster_Config question

Ian Miller. — Thu, 31 Mar 2005 16:03:08 GMT

You may be seeing messages from the other cluster nodes about DTSS. Try
$ MCR SYSMAN SET ENV/CLUS
SYSMAN> DO MCR NCL SHOW DTSS
SYSMAN> DO SHOW LOG NET$DTSS_DISABLE

to see the status of DTSS and the logical.

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 16:21:26 GMT

Hello Ian,
I first rebooted the system as Uwe had suggested and it started the same way as before

I noted down few messages ...

For TCP/IP it often gave errors
however it still continued till is finally reached the

Opcom message 1

Node 0 DTSS
%....OPCOM....%
Message from user SYSTEm on SIZZLE
.....Clock set.....

(I couldn't write down quick enough but it was different than one that was coming before)

The rest two node were giving same message as before

Message from users SYSTEM on XXXX

(Same as above where XXXX = DS25, XP1000)
Event: Too few servers detected from Node..DTSS,

**************************8

I did the commands that you told me and I have the output below (SC4238 is the DS25 and SC1004 is the XP1000)

SYSMAN> DO MCR NCL SHOW DTSS
%SYSMAN-I-OUTPUT, command execution on node SC4238
Node 0 DTSS
at 2005-03-31-16:13:22.181-05:00Iinf
%SYSMAN-I-OUTPUT, command execution on node SC1004
Node 0 DTSS
at 2005-03-31-16:13:21.377-05:00Iinf
%SYSMAN-I-OUTPUT, command execution on node SIZZLE
Node 0 DTSS
at 2005-03-31-16:13:21.315-05:00Iinf
SYSMAN> DO SHOW LOG NET$DTSS_DISABLE
%SYSMAN-I-OUTPUT, command execution on node SC4238
%SHOW-S-NOTRAN, no translation for logical name NET$DTSS_DISABLE
%SYSMAN-I-OUTPUT, command execution on node SC1004
%SHOW-S-NOTRAN, no translation for logical name NET$DTSS_DISABLE
%SYSMAN-I-OUTPUT, command execution on node SIZZLE
%SHOW-S-NOTRAN, no translation for logical name NET$DTSS_DISABLE
SYSMAN>

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 16:24:12 GMT

Just wanted to let you konw I have only done

$ MCR NCL DISABLE DTSS

and not

COPY SYS$COMMON:[SYSMGR]NET$LOGICALS.TEMPLATE SYS$COMMON:[SYSMGR]NET$LOGICALS.COM
Then edit to
$ DEFINE/SYSTEM NET$DISABLE_DTSS 1

because I don't know what it will do to XP1000 (SC1004) or to DS25(SC4238).

How can SC1004 not give problems but this new system can?

I need all the help and any you can provide would be very helpful

Nipun

Re: Cluster_Config question

nipun_2 — Thu, 31 Mar 2005 22:39:10 GMT

Hi Ian and Uwe,

From the messages and little bit of reading it is my conclusion that DTTS although a problem is not responsible for the "blue screen (same as the SRM Conole prompt P000>>>".

Here is why -......

I tried copying the Template file and disabling on the main node (DS25). However when I booted the Satellite node it still gave Opcom message for time for node SC1004 (XP 1000-Satellite node). So I might have to restart XP 1000. I left it at there for now.

But I am wondering whether the above problem would actually be reponsible for the "blue colored screen" just as the beginning
"P00>>> or the SRM console program "

It stays the same. As Uwe said I kept a watch on messages that come prior to error. It seems to show a bunch of "TCP Ip errors" (They went too fast for me to write down) and before that it also says Pagefile.sys not found system initialization continuing..

Then i seem somewhere the Opcom message I mentioned above and then a bunch of things keep going until it stops and then suddenly the Opcam message about DTTS comes and keeps coming. But the moment press enter I go to the $ prompt and everything is present. The only thing it is "blue srm console screen".

And I am lost....

Re: Cluster_Config question

Ian Miller. — Fri, 01 Apr 2005 06:01:49 GMT

as I guess, You are seening the blue srm console becuase DECWindows is not running.

As to if disabling DTSS on the other two nodes will cause a problem, it depends on if you are using it but is unlikely.

to disable it now use DISABLE DTSS as mentioned previously. To prevent it restarting on the next boot define NET$DTSS_DISABLE as described.

Re: Cluster_Config question

Daniel Fernandez Illan — Fri, 01 Apr 2005 06:18:43 GMT

Hi Nipun.

Indeed, it is a message of DTSS client in DECnet phaseV.
In order to disqualify it you can use:
$ncl
NCL>disable dtss
So that one does not activate in the startup you can comment this line:

ENABLE DTSS SET CLOCK TRUE

in file

sys$startup:NET$DTSS_CLERK_STARTUP.NCL

Regards.
Daniel.

Re: Cluster_Config question

Jan van den Ende — Fri, 01 Apr 2005 06:35:32 GMT

Nipun,

I did not yet see your license issue resolved, so I want to come back to it.

You wrote that you entered those PAKs on your system when you did your FIS.

Well, that means they are entered in a file ON A COMPLETELY DIFFERENT DEVICE as when you added tout node to the exiting cluster!!
(barring relocation by logical name, which might become interesting only after you are a lot more familiar with clusters:) the License Database resides on the system disk. But "the system disk" is a different disk in those confirurations!.
So, you HAVE to enter those VMS PAKs again for the other disk!
And then, upon reboot, I expect your Windowing to start up.
If not, we will take it from there.

Success.

Jan

Re: Cluster_Config question

nipun_2 — Fri, 01 Apr 2005 11:07:05 GMT

Hello Thank you all for replying,
I am currently at the following stage

The license I loaded yesterday,but thanks for the explanation Jan that made me more confident. So yesterday I went to the "new node -SIZZLE" and ran the VMS update license and was able to register the

1) OVMS/A smp Asver1200/DS 20 (Product Name :Openvms Alpha)
2) OVMS Alpha Base DS20/DS20E/DS25 LIC (Product Name:OPENVMS-ALPHA

from the "blue screen terminal of SIZZLE".

I also did
$Show license /cluster /unit and I am attaching the file

There were few message before the above two licenses were registered and loaded but not anymore. But I am not sure about Dec Windows license. So I went to node Sizzle

and did show license

Active licesnses on node SIZZLE:

Product UNITS
NET-APP-SUP-150 3150
OPENVMS-ALPHA 75
OPENVMS-ALPHA-ADL 300
PWLMXXXCA07.03 100
Uwe explained yesterday that Dwmotif and other licenses merge so I think Net-APP-Sup-150 might be having those.

Just one more note: I did restart the Satellite node SIZZLE but not the entire cluster. Do I need to do that?

Regarding DTSS I did the exact steps

COPY SYS$COMMON:[SYSMGR]NET$LOGICALS.TEMPLATE SYS$COMMON:[SYSMGR]NET$LOGICALS.COM
Then edit to
$ DEFINE/SYSTEM NET$DISABLE_DTSS 1

So now the Opcom message for DTSS keeps comming but only from SC1004(XP1000-Satellite node).

Some observations:
- I got bunch of TCP IP errors during the booting of Satellite node and somewhere the DTSS message started.

- I was able to run through the cluster_config setup fine (I have attached the file for exact options I have chosen)

one message at end was
********* SHADOWED PAGE/SWAP-DISK WARNING !! ******
**** EDIT THESE PROCEDURES TO INCLUDE LOCAL ****
**** CONFIGURATION COMMANDS NECESSARY FOR ****
**** SHADOWED DISK MOUNTING, PRIOR TO REBOOT ! ****
****************************************************

Final Conclusion:
I understand that DecWindows is not starting up. DTSS I am still reading through comments and trying to make sure whether I have disabled it right.

I had and am assuming that I still have one System disk (on DS25 (SC4238))

If you have any suggestions please continue giving detailed commands so that I atleast know I am doing what you ask me to.

I aplogize for too many postings.

Nipun