Clone between DD Boost devices request for new volumes even when available

Question

Hi,

We have Networker 8.0.2, with two Data Domain systems, one for backups and the another for clones. All this through DD Boost,

The backups work fine, with good transfer rate even some of them are too big, we are talking about some backups are 2 or 3 TB of info. The problem are the Clones, they are on demand, not automatically. When we start a clone job, work fine only few hours, but suddenly stop, without any messages in the details. And when we tried to run the clone job again, we receive the next messages:

"Notice Thursday 9:30:02 AM event media Volume `ddreplicaif.011' ineligible for this operation; Need a different volume from pool `ddreplicaif'"

I don't understan this messages, first, why the clone job isn't available to take the volume that was previously using? And yet, there are another four DD Boost Devices for clones operations, all labeled to the right clone pool, why the clone couldn't take someone of those another four to continue with the job?

I already perform the test of create another new DD Boost device in the Data Domain of replica, to exclude accessibility issues, and when I finish creating the new device, the clone take it and start the clone job. So, anytime that a clone fail I need to create a new device? This is a normal behavior? Why?

Hope someone could give me a light about this.

Best Regards

crazyrov · Answer

I believe the message means you cannot have 2 copies of the same save set on one volume. If you are getting this message it means your clone is completed. But I would suggest you used mminfo to query the savsets that you think have stalled or not completed.

Sent from mobile device please ignore typo

--- Original Message ---

bingo.1 · Answer

I second crazyrov's idea.

We had issues where retried backups would not run sucessfully due to aborted save sets on DDBoost media which seem to confuse NetWorker sometimes. Be aware that such on DDBoost devices will not be deleted right away (which is different compared to AFTDs) but when the next cleanup will take place (sometime within the next 24 hrs). So it is worth to watch for 'suspect' save sets and to delete them manually first.

In this case the DDBoost media - at least for a certain period of time - behaves like a tape. If the SSID has a reference to the media, another one for the same SS is not allowed. Consequently, NW will ask for another media from that pool.

gonzalo.reyes · Answer

Thank`s Crazyrov, Bingo,

Here the thing is that we need to delete the incomplete cloned save sets, right? This is the only way

The problem is that we have communications issues, so some clone Jobs fail and when we try to run the job again, Networker ask for new volumes and this is a little annoying behavior. We will be filling with DataDomain devices, wait for next cleanup job or delete the incompleted savesets in the target DD.

Best Regards

crazyrov · Answer

You can go ahead and delete the incomplete clone saveset manually using nsrmm instead of waiting for NetWorker to clean it up for  you.

PatricJ · Answer

Since the last Networker Updates (NW 8.2.3) i am running in the same issue. And I lack the information how to identify these incomplete savesets to mark them then to delete and manually start the clean run on the data domain device.

How can find this savesets and then manually delete them?

m_makadmeh · Answer

Hello Patric,

You can use the below script to identify the incomplete savesets and delete them from networker:

- Save the script in a file, provide the Target DD Pool Name as an argument/parameter to the script. otherwise, you can 'hard-coded' it in the script.

#!/bin/bash
echo $1 is variable
TOTAL_SSID=`mminfo -xc, -q "incomplete,pool=$1" -r "ssid,cloneid" | grep -v ssid | awk -F"," '{print $1"/"$2}'| wc -l`
x=0

for ID in `mminfo -xc, -q "incomplete,pool=$1" -r "ssid,cloneid" | grep -v ssid |awk -F"," '{print $1"/"$2}'`
do

echo "Deleting Saveset $ID #####"
/usr/sbin/nsrmm -y -d -S $ID

x=$((x+1));
echo -ne "$x of $TOTAL_SSID\r";

done

- (Optionally), SSH to the Data Domain and start the cleaning:

filesys clean status

filesys clean start

filesys clean watch

ble1 · Answer

Patric, I suspect your issue is different and somewhat related to new mdb.  There is only one patch for SP3 out right now (second patch might be out this or next week).  If you suspect that you have broken ssids which are not removed, you should confirm it first. NW should automatically remove it (especially in SP3 since mdb ops are multithreaded); if this is not the case you should open a ticket with support.

bingo.1 · Answer

To check for incomplete save sets on disk, I use the following command:

mminfo -q "family=disk,suspect" -r "client,name,level,savetime(25),ssid,cloneid,sumflags,sumsize,volume"

I do that on a daily base with 2 ideas in mind:

- to find out whether incomplete (and still not deleted) save sets are around at all

- as we use scripted cloning, I want to ensure that I will delete such save sets before the clone jobs are started.

Incomplete save sets will be deleted automatically. However, depending on the disk drive type, a day may pass before this happens. Do delete the save sets instantly, use

- the above command to find the save sets (ssid & cloneid)

- delete each of them with use nsrmm [-y] -d -S ssid[/cloneid]

NetWorker

Clone between DD Boost devices request for new volumes even when available

Was this post helpful?