NDS replication during SP upgrade ??

[rsebasti]

OK,
Will there be NDS replication or synchronization caused by SP upgrades? Am I correct in saying that SP upgrades are only file changes on the servers...
See I plan to remove a faulty RW replica from a remote server. But all the remote servers are on an older SP version which means oder DS nlm version as well. So I thought I should upgrade the remote servers prior to this action...
========================================
I learnt the most
when I stopped talking...

 

[damonuk]

yes, in general SPs are replacement files that have known issues fixed. Novell's knowledgebase is the best place to find what's been fixed.

http://support.novell.com/search/kb_index.jsp

With regards the faulty R/W, have you tried DS repair and checked the comms etc before applying patches? What exactly is the fault? I would try to fix this before doing anything.

 

[salserocito]

I agree with "damonuk", I would run a REAPIR LOCAL DS DATABASE on the problem server until there were no more errors. THEN, upgrade the version of DS to match the rest of your Servers!

Good Luck!

 

[rsebasti]

Running DS Manager this is what i get. Sv2 is the remote server with a RW of the problem replica. I intended to remove Sv2 off the replica ring altogether using dsrepair adv options but I thought I'd update the SP first to be the same level.

Sv1 and Sv3 are on SP5 while Sv4 and Sv2 are on 733. Running dsrepair on Sv2 has the same issues over and over.


- Sv1 - Sv2 - Sv3 - Sv4
Sv1 Mas R/W* R/W Sub
Sv2 Mas R/W R/W Sub
Sv3 Mas R/W R/W Sub
Sv4 Mas R/W R/W Sub

- Sv1 - Sv2 - Sv3 - Sv4
Sv1 Mas R/W R/W Sub
Sv2 Mas R/W R/W Sub
Sv3 Mas R/W* R/W Sub
Sv4 Mas R/W R/W Sub
========================================
I learnt the most
when I stopped talking...

 

[damonuk]

Hi
I assume this is from the Check Synch option in NDS Manager. When did the server last replicate? Have you tried a Check Continuity?
It looks like SVR2 has a problem with communication between it and SVR1 (for replication)

Also, SVR2 also has a problem with the communication between itself and SVR3 (for replication)

Are these -625 errors?

What is happening on SVR2 console (lots of unable to communicate messages?) Is SVR2 time in sync (type the command TIME into the server prompt either at the server or using RCONSOLE or RCONJ etc) If time is not synch'd then try setting the primary time server manually or promoting SVR2 Time type to Primary (both done in Server Options///TIME on the monitor screen)

Are any users having problems?
What speed is the link to SVR2?
Have you tried removing the replica already? Remember that users need a a R/W or Master to authenticate to. If SVR2 is their only local server and you remove the replica, then all login traffic will start crossing the remote link.

 

[rsebasti]

OK,...

It is from the Check Synch option in NDS Manager - YES...

Yes these are -625 errors on SV1 and/or SV3 for the RW replica on SV2...

SV1 and SV3 are in Sydney
SV2 is in Brisbane with a 48K link to Sydney
SV4 is in Melb with a 96K link to Sydney

Time is in Sync - but doing dsrepair and external ref check produces backlink errors and such...

The replica spanning situation is basically not good which is why I am thinking of forcibly removing the replica off SV2 and clearing it with the replica ring. But I thought I'd upgrade the SP on SV2 first to the level of SV1 and SV3. The old admin left it on diff levels...

SV2 has a master replica of the partition which it's users are involved in. But this partition has no issues at all.

The SP upgarde will commence tomorrow prior to planning for the replica removal. During the SP upgrade - will there be heavy NDS replication requirement over the WAN? As I see it - since SP upgrades basically refer to file changes within the server - there shouldn't be any. But I may be wrong which is why I am checking... ========================================
I learnt the most
when I stopped talking...

 

[damonuk]

Right - There shouldn't be heavy NDS traffic during a SP install, EXCEPT as the server reboots and time syncs etc. To be honest I have run different DS levels with no ill effects but it is 'better practice' to have the same DS level.

I would take all replicas off of the Brisbane server except for the 'Brisbane partition'

I would make one of the Sydney servers a R/W of the 'Brisbane partition' or preferably add a second "NDS server" (just for NDS replication, no file and print etc, say a desktop PC) make this new server the master for Brisbane and SVR2 a R/W and have no other Brisbane replicas.

You will really need to sort the -625's out, otherwise you are liable to get immovable objects and partition errors in the NDS; PREVIOUS MOVE IN PROGRESS
to check at the server console type DSREPAIR -a//ADVANCED OPTIONS//CHECK EXTERNAL REFERENCES. The obituary type is 03_INHIBIT_MOVE. Check this TID out

http://support.novell.com/cgi-bin/search/searchtid.cgi?/10012763.htm

If you have the above, you will not be able to modify the affected partition or any of its parents.

Hope this helps.

Damon

 

[rsebasti]

damonuk:
I am getting immovable objects and partition errors in NDS.
When you mentioned:
to check at the server console type DSREPAIR -a//ADVANCED OPTIONS//CHECK EXTERNAL REFERENCES. The obituary type is 03_INHIBIT_MOVE.

Will this help or just verify the problem. If I follow that TID you mentioned will that solve the issue or identify it?

========================================
I learnt the most
when I stopped talking...

 

[damonuk]

Following the TID might solve it, depending on where the objects are moving from and to. I have solved these and also failed to solve these so I cannot say if it will definitely work. If you run through the TID and still have them, then you have only one real option.

Get in touch with Novell and let them remove the obits. They will need a telephone line to a workstation logged in as a supervisor at the branch level where the problem lies (or above of course. From memory Novell use PC Anywhere.

Novell will want the DSRepair logs from all the servers (cleared before running the two procedures they require which I think are CHECK EXTERNAL REFERENCES and CHECK SYNCH but they will tell you.) They will also want the -625 error solved first. Is there any scope for a WAN upgrade? Even to ISDN BRI would help.


It is possible to directly edit the NDS database yourself. IT IS NOT RECOMMENDED. If you make a mistake you could lose your whole tree. Even if Novell cost money they will solve the problem with zero risk to your tree.


Hope this helps.

I am away now until Friday 21:00 GMT.

Damon

 

[rsebasti]

Damonuk:

We proceeded with the SP5 installation with a single glitch to do with policy manager. Users were getting grace logins. TIDS reckon it is to do with NLS.

During the install process, which we haven't really reviewed down to possible glitches, there was a point which required admins password to extend the schema - at which point we couldnt.

A TID mentions NLSLSP5a.EXE which fixes it... but... heres the but

All users including the super of the partition had 0 space on the server with a 0 space limit...

Have you or anyone experienced this and found a workaround? ========================================
I learnt the most
when I stopped talking...