Retention period not being respected

[Malak]

Hello All,

As anyone saw a retention period not being respected?

Netbackup is expiring images before it should.

It’s a User_backup.

CASE: Backup date 29.Mar.2012, retention period 2 Months, expire date 29.Apr.2012

Servers Win2008R2

Client Win2003

Thank you

Malak

[Malak]

Here goes a print screen to help you visualise

[revarooo]

Unusual. What version of NetBackup is the Master server? Have you logged a call with Symantec? If not I would advise that you do.

Is this only happening with user backups? All of them or just this policy?

Does it happen with all retentions or just retentions of 2 months?

What happens if you re-run this policy, does the same issue show?

[Yogesh9881]

i never seen before...

Wha NBU version on master ? are you using Java console or admin console ?

is this backup is disk base or tape base ? is this issue is only with 2 months retension or for all ?

if you opened case with symantec then let us know how they fix this issue.....

[Marianne]

Is it possible that 'someone' has manually changed expiration date with bpexpdate?

I remember a post some time ago where someone has manually changed expiration date but retention level still reflected original value.

The only way to know for sure is if you have logs that would capture bpexpdate. (I guess admin log will contain bpexpdate command(s).)

[mph999]

I'm not aware that NBU has ever done this.  I see / hear of a number of issues like this ... so far I'm not aware of any that have actually been proven to be NBU.

Let me explain why I think it is unlikely.

Everybody who uses NBU uses 'retention' periods, everyone - that's a lot of people.  If NBU was removing images before it should, we'd probably have heard about it by now, and in a nutsell. we haven't.

"CASE: Backup date 29.Mar.2012, retention period 2 Months, expire date 29.Apr.2012"

Intersting that we are exactly one month out

1.  Have the image tapes been imported ?  Is it possible that they have 
2.  If they have been imported, have they been imported back to the same server

In the GUI > Host Properties > Master server under retention periods, how many entries are there for '2 months'

3.  Using this same policy run a backup, what does the retention period get set to.  Considering that you are runnig the same code as you were 'last week' - if it sets the reention correctly, it is reasonable to suggest it did so when these backups were taken.

4.  As Maranne pointed out, in her excellent post, is it possible someone ran bpexpdate ? (Without the log, we'll never know ...)  ... admin log I think

5.  Another possibility is that the policy was set to the wrong retention, some backups ran before this was changed.

When NBU runs a backup it does not 'allocate' the image a retention date.  It allocates it a retention level (1,2,3,4 etc ...) which it 'turns' into a  date  as each retention level represents a date.

For example, if on system A retention peiod 12 is set to be 6 years, and on system B retention period is set to be 6 days ...  If a tape backed up on system A was imported on system B, the retention would change from the expected 6 years to 6 days (which is why I asked if the tapes had been imported).

As far as I know (I've never looked) the retention period (1,2,3,4 etc ) set for an image never changes, even if you use the bpexpdate command.  Therefore, what ever the retention period is set to for an image, is that allocated when the image was backed up.  Remember, the retention period is a number, not a date. It would be useful t determine what this is, for one of the images that as this issue.  For certain it is on the backup header on the tape, but this is tricky to read, it will also probably be in the header file I think (mainly because it can't really be anywhere else) so if you could post up a header file here, it would be useful.

Martin

[mph999]

OK, here is a bit of one of my header files - we see the Retention Level is recorded.

root@womble 1333000000 $ cat womble_ad_disk_1333969447_FULL

VERSION 9

BACKUP_ID womble_1333969447

PROTOCOL_VERSION 10

START_TIME 0

END_TIME 0

SNAP_TIME 0

KBYTES 199872

NUM_FRAGMENTS 1

COPIES 1

CLIENT_TYPE 0

RETENTION_LEVEL 1

THis might be a bit different on your system - this was from a 7.5 system and I 'exported' the 'header file' out of the EMM DB.

So for one of your problem backups, please post up the header file :

/usr/openv/netbackup/db/images/<client name>/<ctime>

( <ctime> is the first four digits of the ctime of the backup id followed by x6 0s  )

Eg.  womble_1333969447

/usr/openv/netbackup/db/images/womble/1333000000/<policyname>_ctime_FULL

If you are at 7.5, you will need to run cat_export command to 'recreate' the header files.

Also post up  output of bpretlevel

Thanks,

Martin

[Malak]

Hello all, thank you for your responses.

After 15 days waiting for Symantec contact, … :(  today we were 2 hours on phone and webex session…

Before all, let me tell you, this is still open an open case…

I’ll try to respond to all of your questions:

Platform is on 7.1.0.3 version (Medias and masters (cluster) servers)

This is not happening only to User backup schedules, I thought so, but I was wrong

I’m using admin console on master server.

For what I could know, no one used bpexpdate on these images; there are a large number of images like this ….

No changes were made in retention level configuration.

None of these images were imported.

These images are on disk.

No changes were made to platform (updates, configurations, etc., in the last months)

So, what we could see today was that, for ALL images already duplicated to tape, we do almost all backups to disk, and then duplicate them to tape, the retention period was on 1month!

The backups we do to disk have, 2 or 3 weeks, and 2 months (between others), all of them that have been duplicated to tape have original retention period (like first backup to disk) but the expiration date is for one month, like the duplication to tape…

We did some tests, and we’ll wait for tomorrow’s duplication to see if it is really as I described before.

Malak

[Jeff_Foglietta]

So all your duplicated images to tape have a retention of one month. Are you overriding the retention period on your duplication to tape in your policy schedule or SLP? Sounds like you are.

[mph999]

Jeff has a good point ...

Now I understand the images are duplicated, there is a possibility that the duplication policy (however you do it) is changing the retention period.

"So, what we could see today was that, for ALL images already duplicated to tape, we do almost all backups to disk, and then duplicate them to tape, the retention period was on 1month!"

The fact that all the images are 1 month is in a way good news, if it was random it would be much more complex.

Lets see what the tests show.

Martin

[Malak]

Hello all,

Symantec support, definitely need to work on their response time, and continuity of cases…

I concluded, not sure if correctly, that the information showed is somehow, incorrect, but all images seem to be there and retention being respected but, like I said, not sure about this yet…

After the backup with 2 months retention to disk, we duplicate it with 1 month retention to tape. It looks like in console reports, it shows the correct retention level, but the expiration date for the one that finishes first..

See below if you understand:

Output from April 24^th:

C:\Program Files\Veritas\NetBackup\bin\admincmd>bpimmedia -client sqlecp -d 03/29/2012 11:40:00 -e 03/29/2012 11:50:00 –U

--------------------------------------------------------------------------------

Backup-ID:            sqlecp_1333017772

Policy:              PL_LOGICA_CLUSTER_SQLECP

Schedule Type:       UBAK

Retention Level      4   <---(2 months)

Number of Files:     1

Compression:         N

Encryption:          N

Image Type:          Regular

Primary Copy:        1

Expires:             11:42 04/29/2012 (1 month) (NOT RIGHT)

  Copy Number:        1

  Fragment Number:    1

  Fragment Size (KB): 11

  Media Type:         Disk

  Density:            -

  File Number:        -

  Offset:             -

  Host:               sre-bkp-med2.bdp.pt

  Device Written On:  -

  MPX:                N

  Expires:            11:42 05/30/2012 (1 month)

  Retention Level:    4    <--- (2 months)

  MediaID:            @aaaad

  Copy Number:        2

  Fragment Number:    1

  Fragment Size (KB): 11

  Media Type:         RMed

  Density:            hcart

  File Number:        65

  Offset:             4563833

  Host:               sre-bkp-med2.bdp.pt

  Device Written On:  9

  MPX:                N

  Expires:            11:42 04/29/2012 (1 month)

  Retention Level:    3    <--- (1 month)

  MediaID:            0918L4

Output from today:

C:\Users\Administrator>bpimmedia -client sqlecp -d 03/29/2012 11:40:00 -e 03/29/2012 11:50:00 -U

--------------------------------------------------------------------------------

Backup-ID:            sqlecp_1333017772

 Policy:              PL_LOGICA_CLUSTER_SQLECP

 Schedule Type:       UBAK

 Retention Level      4    <--- (2 months)

 Number of Files:     1

 Compression:         N

 Encryption:          N

 Image Type:          Regular

 Primary Copy:        1

 Expires:             11:42 05/30/2012

  Copy Number:        1

  Fragment Number:    1

  Fragment Size (KB): 11

  Media Type:         Disk

  Density:            -

  File Number:        -

  Offset:             -

  Host:               sre-bkp-med2.bdp.pt

  Device Written On:  -

  MPX:                N

  Expires:            11:42 05/30/2012

  Retention Level:    4    <--- (2 months)

  MediaID:            @aaaad

Meaning, after the tape image expires, the image information becomes (apparently) OK

[Omar_Villa]

I think you need to be more explicit when you share a problem, after several post I could catch the Netbackup Version you have is 7.3.1, than I only say you have a windows environment but nothing is mention about configuration, but based on the Images you pasted I will assume you are running SLP's because you have a second copy that goes to tape, if this is case why you dont check the image with the nbstlutil <imagename> -U command, it will tell us much more about it, also if you have an appliance or MSDP or any kind of deduplication there is a bug in 7.3.1, but dont know if is your case because there is not enough info about the environment, but check the bug technote:

http://www.symantec.com/business/support/index?page=content&id=TECH172357

let us know

[Jeff_Foglietta]

Can you check your server props? Could it be possible that the retention periods under the server properties were modified and the value for retention period 4 is now 1 month and not 2?

[Malak · Accepted solution]

Hello All,

It looks like this is "normal behavior" on NB, it show the expiration date on the image that will expire first, so, if you have a backup with 2 month retention period, and you duplicate an image with less time, you will see that you have a backup with 2months retention but the expiration date will be less, igual to 2nd copy.

example:

backup today (Jun 1st) with 2 month retention period, expiration date will be Aug 1st, right?
then you duplicate, and 2nd copy has 1 month retention, this will be July 1st, right?

When checking for this system backups, you will see that you have a backup with 2 month retention, but expiration date is July 1st...

according to Symantec.... "normal behavior" ...

Thank you all.
Be well.

Malak

[Jeff_Foglietta]

Something doesn't sound right here.. The expiration should reflect that of the primary copy or with multiple copies reflect the expiration of each copy.