Veritas-bu

[Veritas-bu] question about incremental vs full on multpile policies

2005-01-07 08:31:34
Subject: [Veritas-bu] question about incremental vs full on multpile policies
From: dave.markham AT fjserv DOT net (Dave Markham)
Date: Fri, 07 Jan 2005 13:31:34 +0000
Dave Markham wrote:

> This is great and i ran ITC last night after setting the multple 
> copies section in the each schedule.
>
> I have cope this morning and all backups have worked but when i try 
> the bpduplicate command it says
>
> INF - Skiing backup id <hostname>_<id>, it already has 2 copies
>
>
> Any ideas how i can identify tapes used for the second copy last night?
>
> Cheers
> Dave
>
>
>
> Tim Hoke wrote:
>
>> Ok, Let me boil this down and see if I've got it right.
>>
>> Basically, you want two copies of data, need to identify which tapes 
>> contain that second copy (or first, doesn't really matter) and be 
>> able to run a script (i.e. set of commands) to identify which tapes 
>> an opperator should eject and send offsite?
>>
>> This to me sounds like vault.
>>
>> Now, there's no reason you can't write your own vault, it's just more 
>> elegent with the built in feature.
>>
>> So, assuming my analysis is correct, lemme throw this out to you.
>>
>> 1. Use ITC and direct copy 2 to the "Offsite" Pool. This creates your 
>> second copy and doesn't mess with the archive bits or have problems 
>> with missing files based on time stamps.
>>
>> 2. Now, identify which tapes were used that need to be ejected:
>> bpduplicate -p -cn 2 -hoursago 18|grep ^Media |awk '{print $4}'
>>
>> As far as changing primary copy, it's really not all that hard.  It's 
>> as simple as using the bpduplicate -npc command.  It's a very common 
>> thing that people run.
>>
>> -Tim
>>
>> On Wed, 5 Jan 2005, Dave Markham wrote:
>>
>>  
>>
>>> Tim Hoke wrote:
>>>
>>>   
>>>
>>>> I'm going to avoid answering your question and ask a couple of my own.
>>>>
>>>> What version of NBU are you using?  Is there any chance you can 
>>>> make use of Inline Tape Copy?  It would allow you to make two 
>>>> copies of the data in a single backup cycle.  It also allows you to 
>>>> place the "duplicate" in a different pool.
>>>>
>>>> HTH
>>>> -Tim
>>>>
>>>> On Wed, 5 Jan 2005, Dave Markham wrote:
>>>>
>>>>
>>>>
>>>>     
>>>>
>>>>> Ok,
>>>>>
>>>>> I am running Netbackup on solaris 9 with 2 media servers about 22 
>>>>> unix clients and 7 windows200 clients.
>>>>>
>>>>> I have normal backups which run daily. ( Cinc ) and full backups 
>>>>> at weekends.
>>>>>
>>>>> I have decided to copy my policies to Offsite_<policy> and change 
>>>>> the volume pool on each schedule to be my offsite tape pools for 
>>>>> cinc and full.
>>>>>
>>>>> Now my question is as all files, and clients are the same on the 
>>>>> normal policy and my offsite policy what happens with the 
>>>>> incrementals, i.e where do they take there times from for being 
>>>>> incremental files.
>>>>>
>>>>> for e.g
>>>>>
>>>>> Pol1 ( normal ) full backup on sat
>>>>> Pol1 ( normal ) inc on sunday will get files changed since saturday
>>>>>
>>>>> OffPol1 ( offsite ) full backup on sat
>>>>> OffPol1 ( offsite ) inc on sunday . Will this get files changed 
>>>>> sine OffPol1 full backup or could it maybe miss some out as they 
>>>>> could be reset from Pol1 (normal) inc ???
>>>>>
>>>>> That make sense?
>>>>>
>>>>> Im a bit confused cant find much docs to explain it
>>>>>
>>>>> Cheers
>>>>>
>>>>> Dave
>>>>> _______________________________________________
>>>>> Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
>>>>> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>>>>>
>>>>>  
>>>>>       
>>>>
>>>>
>>>>
>>>>     
>>>
>>> Yeah im using 5.0Mp3
>>>
>>> I have posted a few mails regarding using duplication, vs inline 
>>> tape copy on here.
>>> The problem is i need to identify each day which tapes were used for 
>>> offsite and auto mail the operators to take those tapes out.
>>>
>>> The way i can easily do it with using a different policy and 
>>> schedule for offsite is by using bpverify -p -sl offsite_inc 
>>> -hoursago 18. I launch this from a script which wraps bits n bops 
>>> around it and checks no jobs are still running or sleeps.
>>> This then mails ops and they know which tapes to take out and i dont 
>>> need to worry about it unless something goes wrong. I couldnt work 
>>> how to identify tapes easily with the duplicate or inline tape copy 
>>> method.
>>> I could easily work out all tapes in the robot which are in the 
>>> offsite pool so could identify that way, but then was wondering how 
>>> to do a restore from these offsite tapes being a second copy.
>>> I would need to change the copy netbackup tries to use for restore 
>>> which gets messy, where as this way all these images are separate 
>>> and so i just put offsite tapes in and select the image to restore from
>>>
>>> Make sense?
>>>
>>> Im just concerned about what gets backed up and incremental bits 
>>> with 2 inc's running.
>>>
>>> Dave
>>>
>>>   
>>
>>
>>
>>
>>  
>>
>
> _______________________________________________
> Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>
I have been trying some alternate commands and this seems to do what i 
require, although im not sure if its as good as the duplicate one which 
isnt working for me :-

bpimagelist -hoursago 14 | awk '$2==2 {print $9}' |sort | uniq


Dave

<Prev in Thread] Current Thread [Next in Thread>