This post is more than 5 years old
6 Posts
0
8699
backup of clients work, but no index entry for the saveset is created
This problem just started occurring on a couple dozen Networker clients (out of 325 total clients). Networker server is v8.1.1.2 build 280, Networker Console server is 8.1.1.7. All Networker system components are virtual machines (server, console, proxies).
The first thing I noticed appeared in the job logs. Although the job completes with a successful status, at the end of the log I get this "Unable to find any full backups of the save set 'abcxyz.local: *FULL*' in the media database. Performing a full backup.
I then ran some queries via mminfo, and then went to the GUI (media tab, savesets) to see if any savesets were actually captured. Good news is the savesets are being created. However, there is no index entry for the saveset, so I can't run a restore as NW has no entry in the client index?? What's up with that??
First I need to figure out why the savesets never make it to the client index, and then I need to rebuild the index for the client.
I tried running a nsrck with various levels, but had no success populating the index with the saveset info. At this point I am afraid to make things worse.
I suspect this is related to some Active Directory work that was done last week, and since Active Directory provides all my DNS services, I'm thinking there is a connection, but "$nslookup" works just fine, both forward and reverse.
Any ideas from the community would be appreciated.
Mark_Zurl
6 Posts
0
August 29th, 2014 12:00
I contacted technical support who provided me a solution. I believe the problem has been resolved as I've run a test and it seems the problem has disappeared.
This problem appears to have been created by the proxy used in the NW client backup job. The proxy I used for the backup had in the "Apps & Modules" section this line:
VADP_DISABLE_FLR=YES
As explained to me by technical support, when the NW client job starts, it inherits this definition. For those unfamiliar with VADP_DISABLE_FLR, this configurable parameter tells Networker to disable individual file level recovery from an image backup.
In this case I had written a job to create an image level backup of my proxy so that should I ever lose the proxy, I could restore it. However, any subsequent image backup of any Networker client (that uses that proxy) inherits this parameter, so NO index for individual files is created; hence no index could be found.
Many thanks to Dan Webster from EMC Networker technical support for finding the solution to this difficult problem I had.
crazyrov
4 Operator
4 Operator
•
1.3K Posts
0
August 26th, 2014 05:00
Mark,
First you will have to ensure there are no duplicate client id for the same client, when this happens the client index is purged ever time the nsrim runs (nsrim runs automatically once a day). run the following command on your backup server.
mminfo -avot -q client=client_name -r clientid | sort | uniq
bingo.1
2.4K Posts
0
August 26th, 2014 06:00
If I understood you correctly, the backup exists. So there must be an entry in the media index.
It might be 'misplaced' though due to hostname/client id issues. This is when NW will must start a new backup cycle with a FULL.
But you also say...
"However, there is no index entry for the saveset, so I can't run a restore as NW has no entry in the client index??"
This is contradactionary as you refer to the 'index entry for the save set' (pointing to the media index) but also mention the missing file index infomation.
Don't be afraid - worst case you can still recover the whole save set without the file index.
As the file index info is written during the backup, it must exist when the backup was successful.
However, the file index backup at the end of the backup could be missing.
So i am not really sure what your real issue is. Please clarify.
Mark_Zurl
6 Posts
0
August 26th, 2014 08:00
I executed the mminfo to verify that duplicate clientid's do not exist. No duplicates were found.
To clarify, a backup job runs, completes successfully but logs "Unable to find any full backups of the saveset"
Within the NW Administration GUI, if I go to the media tab (left pane) and select "savesets", I can successfully query all the savesets for the clients in question. However, I cannot restore from the savesets. If I try to restore from either the NW Admin GUI, or go to the proxy and run NW User to run a restore, all you get is "no index found".
As I understand it, the media database contains the savesets, and the index (should) contain the pointers to the savesets in the media database. If the index contains no entries to the saveset then you will get exactly what I am seeing when I try to restore, e.g., "no index found".
As near as I can tell, a backup job runs, doesn't see any previous backup so runs a FULL, then stores the saveset but does not correctly update the client index.
Thanks for everyones response so far.
Mark Z
crazyrov
4 Operator
4 Operator
•
1.3K Posts
0
August 26th, 2014 08:00
Mark,
Have you tried browsing the backups soon after completion of a successful backup?
Mark_Zurl
6 Posts
0
August 26th, 2014 08:00
crazyrov -
When you say browse after completing, I assume you mean attempt a restore. Immediately after the backup completes, any attempt at restore generates a "no index found".
I can execute an " mminfo -avot -q client=xyzabc.local " to verify the saveset exists, but any attempt at restore only generates a no index found?? If Networker can find the saveset via mminfo, why would there not be an index entry?
ble1
2 Intern
2 Intern
•
14.3K Posts
0
August 26th, 2014 09:00
Is this VM (VADP or VBA) backup or traditional one (in guest)?
Mark_Zurl
6 Posts
0
August 26th, 2014 09:00
Note:
All of the previous backups I referred to in this case were image backups, e.g., "nsrvadp_save".
I then ran a test. I performed a traditional file system backup on one client that was having the problem. After the backup completed, I could restore from that saveset, no problem.
It appears then this problem is related to virtual machine image backups. A traditional filesystem backup works A-OK and the index gets updated correctly.
ble1
2 Intern
2 Intern
•
14.3K Posts
0
August 26th, 2014 15:00
Not sure how it is in 8.1, but in 8.0 image backup could only be restored as ssid restore (meaning it was not browsable). You should try instead VBA with 8.1 which is now what is being pushed by both EMC and VMware.
Mark_Zurl
6 Posts
0
August 27th, 2014 05:00
Update:
I have now contacted technical support. When I resolve this I will post the solution as it may help someone in the future.
The problem appears to be that "nsrvadp_save" images do not make it into the clients index, but a traditional filesystem backup "save" does. Any restore of individual files from the image backup cannot be performed, and generates an error "no index found". The error appears when trying to restore from either from the NW GUI or a proxy. At this time I have not tried the restore from the command line, but I would think that would fail to.
Thanks for everyone's input so far.
bingo.1
2.4K Posts
0
August 27th, 2014 05:00
Your expections sound logical because: "You can only recover what has been backed up."
The same story goes for the index: "You can only use it (no matter which way) if it has been created."
BTW - it would have been much easier if you would have pointed out in the beginning that these are not only VMs but that you also used VADP.