This post is more than 5 years old
61 Posts
0
1577
Networker Cloned inc backups stay browsable on tape.
Hello!
I wrote about this issue some time ago and got some good feedback that led me to a diagnosis. I'm hoping for a solution!
The issue is as follows. my tapes don't expire as they should.
The pertinent parts of the backup infrastructure are as follows.
one zone
a backup server in location A
a Data domain in location A (DDA)
a storage node with a tape library in location B
a Data domain in location B (DDB)
Backups from DDA a are clone controlled replicated to DDB. all backups from DDB, including the cloned backups from DDA are cloned off to a tape library.
When browsing tapes that should be recyclable for over a month, i see a ton of incremental backups that are browsable, but all of the fulls and SOME of the incremental are marked as recyclable.
i cant seem to find any rhyme or reason to this. the original save sets are long gone on our DDA and DDB, so im not sure why the incrementals are still "needed".
The best solution i can think of is to alter the clone to tape job to only include full backups, but that is not optimal.
does anyone have any input they can give me on this?
NetworkSupport-
61 Posts
0
October 11th, 2014 04:00
Well, this is what i have found.
the systems are keying off of their individual schedules. This is very odd... i understood that the group scheudle was supposed to override. and for SOME clinets it does... 99% of them had a "default" schedule. so after i individually changed them to the correct schedule they all have fallen into line.
That combined with synthetic full backups was causing the "eternal incremental" issues on my tapes.
i know a patch was just released recently, mayhaps its time to update!
NetworkSupport-
61 Posts
0
September 27th, 2014 19:00
here is an interesting piece of info......
it didnt become apparent until i ran that mminfo. there are no full backups of the vm's in the tapes... or the data domain for that matter. the backup group is actually set to explicitly backup fulls on saturday. which it still is on my side of the world. Im going to attempt to set the members of the group for a saturday full backup and see what that does.
the "force incremental" was checked. after unchecking it, it still runs an incremental....
ill post more in a bit.
NetworkSupport-
61 Posts
0
September 27th, 2014 19:00
ill check on this now and see if it sheds any light.
thanks for the quick response!
ble1
2 Intern
2 Intern
•
14.3K Posts
0
September 27th, 2014 19:00
Well, it is 4:15am here and I just completed rollback after not so successful SAP upgrade and don't feel sleepy
I usually only trust mminfo. Of course, I do not say you won't perhaps find some wrong things like sequence which should have expired as there were couple of bugs with that in the past (usually with DDB devices), but to get to the bottom of your problem best thing is to check this first.
ble1
2 Intern
2 Intern
•
14.3K Posts
1
September 27th, 2014 19:00
You can do following:
mminfo -avot -q client= -N -r 'name,client,ssflags,sumflags,clflags,savetime(20),ssbrowse(20),ssretent(20),clretent(20),level,pool,volume'
and you compare what you get for specific chain. Above output will be listed according to time sequence of savetime. You can omit -N part if you wish to list it for whole client.
NetworkSupport-
61 Posts
0
September 27th, 2014 19:00
sorry to ask for baby steps here, but mminfo dosent like the query. its reporting that there is an "unknown report constraint: 'name. when i remove name its the next one in the list. thoughts?
NetworkSupport-
61 Posts
0
September 27th, 2014 19:00
ya, something happend in the copy-paste. a manual typing worked
. thanks again!
i think i may see something. ill post back in a moment!
ble1
2 Intern
2 Intern
•
14.3K Posts
0
September 27th, 2014 19:00
Strange as it works for me. Error says something is wrong with syntax after -r and it seems to fails on first one - not sure why in your error it says name. as it should be name,
NetworkSupport-
61 Posts
0
September 27th, 2014 20:00
so it looks like.. the Full Tuesday, Wednesday, Thursday and Saturday schedules are not "in effect". none of the VM's associated with those backups are following the plan. inc every day except the day specified....
is it possible for the schedules to be "corrupted" some how?
ble1
2 Intern
2 Intern
•
14.3K Posts
0
September 27th, 2014 21:00
Where do you set schedule?
NetworkSupport-
61 Posts
0
September 27th, 2014 21:00
in the gui.
configuration>schedules
i then apply them to the GROUPS as opposed to the individual systems. however when it still wouldn't work, i applied it to the systems.
That still didn't work and i had to set the SAVE LEVEL of the GROUP to FULL rather than the schedule. that got me a full backup. its doing one now.
further investigation showed that there were other groups using the listed schedules that DID have full backups on the appropriate days.... it is isolated to 4 VADP backup groups. im going to do some further trial and error to see if i can get them to do a full backup on other days...
any additional insight is appreciated
NetworkSupport-
61 Posts
0
September 28th, 2014 14:00
at this time i SUSPECT my issue is that i have synthetic full backup checked. i still need to run some example tests to verify that is the case, but from what im reading that makes sense.
i will update with results later.
NetworkSupport-
61 Posts
0
September 30th, 2014 05:00
ill have to chime back in on this in a few days. the schedules are in place, and i have disabled synthetic full backups. provided i go through the whole week without issue, i will report back that was the "issue".