Networker savegrp and mminfo inconsistencies

Question

Since we now have DPA which checks backup success using mminfo and a mechanism to alert on backup failures using savegrp we can now check backup results using both methods and hope they should match. The two reports show discrepancies, when digging deeper into the discrepancies we have found that mminfo and savegrp do not always match here is an example of just one case:

(NW 7.3.4 (I know EOL...!))

09/16/09 20:15:27 savegrp: command ' save -s elonsaplgto2-cbs -g gbwmappropc03 -y "Sat Jan 16 20:09:58 GMT+0000 2010" -LL -f - -m gbwmappropc03 -o VSS:*=off -t

1253042111 -l incr -q -W 78 -N "SYSTEM DB:\\" "SYSTEM DB:\\"' for client gbwmappropc03 exited with return code -1

09/16/09 20:15:27 savegrp: job (2445155) host: gbwmappropc03 savepoint: SYSTEM DB:\ had ERROR indication(s) at completion.

* gbwmappropc03:SYSTEM DB:\ 1 retry attempted

* gbwmappropc03:SYSTEM DB:\ Removable Storage Database - rsmow: Unable to open a session: 0x00000015.

* gbwmappropc03:SYSTEM DB:\ Removable Storage Database - Error saving

* gbwmappropc03:SYSTEM DB:\ 39078:save: RPC error: RPC cannot encode arguments

* gbwmappropc03:SYSTEM DB:\

* gbwmappropc03:SYSTEM DB:\ 7162:save: save of SYSTEM DB:\ to elonsaplgto2-cbs failed

09/16/09 20:15:27 savegrp: gbwmappropc03:SYSTEM DB:\ failed.

09/16/09 20:17:18 savegrp: job (2445035) host: gbwmappropc03 savepoint: C:\ had WARNING indication(s) at completion.

20090916 20:18:38 Failed: savegrp -I -l inc -y 4months -G gbwmappropc03

But mminfo shows success:

mminfo -av -q "client=gbwmappropc03,savetime>09/16/09" -r "name,savetime(22),client,ssflags,sumflags"

name date time client ssflags fl

SYSTEM STATE:\ 09/16/09 20:11:46 gbwmappropc03 vF cb

D:\ 09/16/09 20:14:16 gbwmappropc03 vF cb

C:\ 09/16/09 20:13:44 gbwmappropc03 vF cb

ASR:\ 09/16/09 20:11:24 gbwmappropc03 vF cb

SYSTEM DB:\ 09/16/09 20:14:18 gbwmappropc03 vF cb

SYSTEM DB:\ 09/16/09 20:12:20 gbwmappropc03 vF cb

This is bad news since we are now longer confident that if savegrp says we have a success or failure it is reporting the truth.

We have seen examples of savegrp reporting success and mminfo failure and the reverse, worse case savegrp reports success then you can t do a restore because mminfo disagrees.

Has anyone else seen this?, not an easy one to spot unless you have a reporting mechanism like the above or you go round doing mminfo all the time.

ecarter11 · Answer

Hi Pierre,

As the marketing guy I'm not going to be able to give you a resolution here - but I did want to say thanks for the post. You bring up something that would concern me too. One of the first things that struck me was that in 7.3 there was a way to configure NW so Savegroup wouldn't show you "failed" for benign errors (like open file issues on a marketing server!). With that I could see savegroup report showing something different than mminfo - but perhaps not the other way around? I think some folks on here will take a look at what you've pasted in and offer a hand. Is this something our support guys are working?

Regards and thanks!

-Eric

PierreML · Answer

Hi Eric, Yes I have a case open with support to look at this issue. Regards Pierre

sweller1 · Answer

Thank Pierre,

We will work with Support to ensure this is given the proper attention and escalate if needed.

Beyond this specific issue I was hoping to get some thoughts from you on the recent enhancments and current plans you may be aware of around NetWorker. We have been focusing on ease of use and enhancing the administrators experience as well as a few other areas. What kind of things would you like to see in NetWorker going forward?

Thank you very much.

-Steve

PierreML · Answer

Hi Steve,

Our number one wish it for a new feature called Storage node parallelism which would load balance backups across multiple storage nodes, a task which currently has to be done manually. This was discussed at the EMC TAB and Richard Reitmeyer is aware.

Regards

Pierre

AHP1 · Answer

Sorry for coming in a bit late on this but I have also seen this issue.

I noticed this as I create reports from the csv files you can create using gstclreport which is the command line interface into the nmconsole reports.

People were quering my stats as mminfo was showing completed backups that didn't appear on my report. Cross checking against the output of daemon.log showed that gstclreport was correct and mminfo was wrong.

Regards Andy

NetWorker

Networker savegrp and mminfo inconsistencies

Was this post helpful?