Extraordinarily Long Reboots

[bopritchard]

In the past couple weeks my netware server has suffered 3 abends in the f-prot process...i was able to down the server succesfully...but when i attempted to bring it back up it took upwards of 45 minutes for it to boot...system console would come up then it would sit forever...finally the messages Checking Subdirectories...etc, would appear and it would come online...

what in the world is going on...this has just started recently...normal reboots would be 2 to 3 minutes max

 

[Jontmke]

When it pauses, or is just sitting there, is it mounting a volume? Sounds like it with that "Checking Subdirectories" message. What I'm getting at is you need to determine what is the server doing when it is taking so long so that you can fix the problem.
Is it loading an NLM? Mounting a Volume? A device driver?
Look on the screen and follow what is loading to see when it slows down.
If it is the Volume mount taking that long, I would do a vrepair.

Jon

There is much pleasure to be gained from useless knowledge. (Bertrand Russell)

 

[TheLad]

Sounds like it is mounting a volume during the pauses. It isn't uncommon for servers to take longer than the norm to mount volumes after an abend. I don't know why and there doesn't appear to be a TID about it, it's just something I have noticed.

The volume isn't being repaired automatically using VREPAIR is it?

-----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[bopritchard]

how would i know whether vrepair is being run automatically?...is that a startup setting?

 

[Jontmke]

No, it only runs if it notices a problem. Like scandisk for Netware.
Watch the server boot and look at the messages that come across the screen. If another screen comes up, switch back to the main screen, (ctlr-esc) to see what is going on. It will tell you what is loading. Along the same lines, scroll through the screens that have loaded to see if they are having issues.

Jon

There is much pleasure to be gained from useless knowledge. (Bertrand Russell)

 

[Lou0686]

Hi,

I don't believe that the volume is repairing itself because you would automatically see the Vrepair screen.

At this point I would try to identify at what stage the problem is occuring by loading the stages manually. You can do this like so:

1. At the C:\NWSERVER prompt, type "Server -na -ns"

2. Next type "startup"

3. Next "loadstage 1", then 2 and 3, etc...

Stop at the stage that takes a long time.

4. Next "list stage #". This will tell you what is loading at the each stage.

5. Reboot, start again but this time when you get to the affected stage, load the modules individually and in the order listed.

This should get you a little closer to what is happening.

Lou

 

[bopritchard]

i went through each of the load stages fine...no delays...so then i ran autoexec...got to

NLSTRAP: Server Name is ICFP1
NLSTRAP: Community String set to public
NLSTRAP: Time Interval set to 10

and everything stops...sits there for 45 minutes...doesn't nls have something to do with licensing?

 

[TheLad]

I would comment out the MOUNT ALL line from AUTOEXEC.NCF and see if the server takes as long. If it does not, type MOUNT ALL at the console and see if it definately is the volumes.

If not, have you run DSREPAIR to repair your database?

-----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[Lou0686]

Because NLSTRAP was the last item you saw does not mean it is the cause but rather whatever is happening afterwards.

As my next step I would remark out as much as possible from the Autoexec, restart the server and manually load everything that was remarked out, one at a time with about 30 seconds or so between each one.

Lou

 

[marvhuffaker]

I've seen servers hang at that point in the following scenarios: 1) Directory is corrupt 2) Server can't communicate with other servers for some reason.

See if you can get to a prompt without loading all the autoexec stuff..

SERVER -NA

If you can get to a prompt, and SYS volume has mounted, try running a DSREpair on the database.

There are several TIDs that cover this type of problem, you should be able to search novells knowledgebase and come up with something.

Marvin



Marvin Huffaker MCNE, CNE
Marvin Huffaker Consulting

http://www.redjuju.com