Conference decwet::networker

426.0. "nsrindexd and Dr. Watson on NT Alpha 4.0" by DIBDIB::DBATES (Time for a note) Thu Feb 20 1997 02:48

Our NSR main server hasd started giving Dr. Watson errors with the nsrindexd
process being the application that gives the error.
As a result none of our backups are working and we cannot connect to the server.

Unfortunatley our NSR expert is away on holiday and I am not overly
familiar with the product.

The system is an Alpha running NT server 4.0.

I have had a looked at the daemon.log and this is what the end of it looks like.

2/20/97  9:53:47 nsrd: index notice: Check failed for client ntonly.reo.dec.com (error -5)
 2/20/97  9:53:47 nsrd: index notice: completed checking 5 clients
 2/20/97  9:53:49 nsrmmdbd: Starting compression of media database
 2/20/97  9:53:58 nsrmmdbd: Finished compression of media database
 2/20/97  9:53:59 nsrd: index notice: nsrim has finished cross checking the media db
 2/20/97  9:54:02 nsrindexd: Rebuilding btrees...
 2/20/97  9:55:47 nsrd: nsrindexd has exited with status 0
 2/20/97  9:55:47 nsrd: shutting down
svcipx_create: socket creation problem: Address family not supported by protocol family
 2/20/97  9:57:52 nsrd: server notice: started
 2/20/97  9:58:14 nsrmmdbd: media db is cross checking the save sets
 2/20/97  9:58:15 nsrmmdbd: media db is open for business
 2/20/97  9:58:16 nsrindexd: checking index for wade.reo.dec.com
 2/20/97  9:58:16 nsrindexd: checking index for offsvr.reo.dec.com
 2/20/97  9:58:16 nsrindexd: checking index for samedi.reo.dec.com
 2/20/97  9:58:16 nsrindexd: checking index for ntpieces.reo.dec.com
 2/20/97  9:58:17 nsrindexd: checking index for ntonly.reo.dec.com
 2/20/97  9:58:17 nsrindexd: Checking volume directory
 2/20/97  9:58:17 nsrindexd: 	FATAL file saverecs.x0 has bad first page
 2/20/97  9:58:17 nsrindexd: 	FATAL Can't remove file saverecs.x0
 2/20/97  9:58:17 nsrindexd: Cannot mount index F:\nsr\index\ntonly.reo.dec.com\db (-5)
 2/20/97  9:58:17 nsrindexd: Checking volume F:\nsr\index\ntonly.reo.dec.com\db
 2/20/97  9:58:17 nsrindexd: 	Volume header ok
 2/20/97  9:58:17 nsrindexd: Checking volume directory
 2/20/97  9:58:17 nsrindexd: 	FATAL file saverecs.x0 has bad first page
 2/20/97  9:58:17 nsrindexd: 	FATAL Can't remove file saverecs.x0
 2/20/97  9:58:17 nsrindexd: Cannot mount index F:\nsr\index\ntonly.reo.dec.com\db (-5)
 2/20/97  9:58:17 nsrindexd: Check failed for client ntonly.reo.dec.com (error -5)
 2/20/97  9:58:17 nsrindexd: completed checking 5 clients
 2/20/97  9:58:22 nsrd: index notice: Check failed for client ntonly.reo.dec.com (error -5)
 2/20/97  9:58:22 nsrd: index notice: completed checking 5 clients
 2/20/97  9:58:24 nsrmmdbd: Starting compression of media database
 2/20/97  9:58:34 nsrmmdbd: Finished compression of media database
 2/20/97  9:58:34 nsrd: index notice: nsrim has finished cross checking the media db
 2/20/97  9:58:38 nsrindexd: Rebuilding btrees...
 2/20/97 10:19:22 nsrd: nsrindexd has exited with status 4
 2/20/97 10:19:22 nsrd: shutting down
 2/20/97 10:19:46 nsrd: successful shutdown

Could someone please suggest what we should do to recover our NSR system

Many thanks

David

First, please save for engineering:
	nsr/logs/*
	nsr/index/ntonly

This is a problem we've seen at a customer site and
they've been unable to provide us with adequate
data.

Next, to get you back in operation, how do you
feel about not being able to on-line browse the
old ntonly backups?  If this is not a problem,
the easiest thing to do is to rename the index
directory for ntonly (nsr/index/ntonly -> 
nsr/index/ntonly_saved_for_engineering), then
restart the server.  This will create a new
empty index for ntonly.  If you need to recover
files from the old ntonly backups, you'll have
to use the scanner command on the volumes 
required.

Next, tell me how I can ftp the logs and index
out of the server system so I can analyze them.

Thanks/Mark

Well we have used mmrecov with the last successful backup.
Our backups are working again, but I think there may still be
a problem with NTONLY.
I think that mmrecov has fixed the volume index/database
but maybe not the client indexes.

I will try your suggestion for recreating the NTONLY index, once the
backups finish.

Thanks

David