Re: Archival backup problems

2005-06-23 Thread Kevin Dalley
Jon LaBadie <[EMAIL PROTECTED]> writes:

> On Tue, Jun 21, 2005 at 02:11:34PM -0700, Kevin Dalley wrote:
>> 
>> 
>> Matt Hyclak <[EMAIL PROTECTED]> writes:
>> 
>> > On Tue, Jun 21, 2005 at 01:26:49PM -0700, Kevin Dalley enlightened us:
>> >> Thanks for the suggestion.  Setting record to no seems like a good idea.
>> >> 
>> >> However, dumpcycle set to 0 doesn't work for me.  amdump tries to run
>> >> a full backup each day.  I only use 1 or 2 tapes per amdump, and I
>> >> need 6-10 tapes for a complete archival backup.  So I need a few days
>> >> to complete an archival backup.  Setting dumpcycle to 0 seems force
>> >> the full backups to start all over again each day, which means that
>> >> DLEs are repeated.
>> >>
>> >
>> > Do you not have enough holding disk space to hold the entire dump set? You
>> > didn't say how big your tapes were, or how much data was getting dumped. If
>> > you have enough holding space, you can run the amdump, let it spool to
>> > holding disk, then just amflush until you've got everything to tape.
>> >
>> > The other option is that amdump can take hostnames as options, so you could
>> > run the job with a select number of hosts over a few day period. 
>> 
>> 
>> No.  I don't have enough disk space to hold the entire dump at once.
>> Sorry for not mentioning it.  There's enough space for the daily
>> amdump, but not for a complete archival dump.
>> 
>> Adding the hostname (and disk for the Windows machines) should solve
>> my problem.  Thanks for the suggestion.
>> 
>
> Another possibility might be to do similar to your original
> arrangement, long dumpcycle, no increments, then run amadmin
> to force level 0 backups of specific DLE's the next run.
> Then do the archival run for just those.  Repeat as needed.


Thanks.

My archival backups seem to work now.  I did set up a long dumpcycle
(178 days).  That will remind me when I need to start my next archival
round of dumps.  I'm not explicitly skipping incrementals now.  It is
a pain having to run just one host at a time.  The backups are
definitely slower than they were.

-- 
Kevin Dalley
[EMAIL PROTECTED]


Re: Archival backup problems

2005-06-21 Thread Jon LaBadie
On Tue, Jun 21, 2005 at 02:11:34PM -0700, Kevin Dalley wrote:
> 
> 
> Matt Hyclak <[EMAIL PROTECTED]> writes:
> 
> > On Tue, Jun 21, 2005 at 01:26:49PM -0700, Kevin Dalley enlightened us:
> >> Thanks for the suggestion.  Setting record to no seems like a good idea.
> >> 
> >> However, dumpcycle set to 0 doesn't work for me.  amdump tries to run
> >> a full backup each day.  I only use 1 or 2 tapes per amdump, and I
> >> need 6-10 tapes for a complete archival backup.  So I need a few days
> >> to complete an archival backup.  Setting dumpcycle to 0 seems force
> >> the full backups to start all over again each day, which means that
> >> DLEs are repeated.
> >>
> >
> > Do you not have enough holding disk space to hold the entire dump set? You
> > didn't say how big your tapes were, or how much data was getting dumped. If
> > you have enough holding space, you can run the amdump, let it spool to
> > holding disk, then just amflush until you've got everything to tape.
> >
> > The other option is that amdump can take hostnames as options, so you could
> > run the job with a select number of hosts over a few day period. 
> 
> 
> No.  I don't have enough disk space to hold the entire dump at once.
> Sorry for not mentioning it.  There's enough space for the daily
> amdump, but not for a complete archival dump.
> 
> Adding the hostname (and disk for the Windows machines) should solve
> my problem.  Thanks for the suggestion.
> 

Another possibility might be to do similar to your original
arrangement, long dumpcycle, no increments, then run amadmin
to force level 0 backups of specific DLE's the next run.
Then do the archival run for just those.  Repeat as needed.

-- 
Jon H. LaBadie  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road(609) 252-0159
 Princeton, NJ  08540-4322  (609) 683-7220 (fax)


Re: Archival backup problems

2005-06-21 Thread Kevin Dalley


Matt Hyclak <[EMAIL PROTECTED]> writes:

> On Tue, Jun 21, 2005 at 01:26:49PM -0700, Kevin Dalley enlightened us:
>> Thanks for the suggestion.  Setting record to no seems like a good idea.
>> 
>> However, dumpcycle set to 0 doesn't work for me.  amdump tries to run
>> a full backup each day.  I only use 1 or 2 tapes per amdump, and I
>> need 6-10 tapes for a complete archival backup.  So I need a few days
>> to complete an archival backup.  Setting dumpcycle to 0 seems force
>> the full backups to start all over again each day, which means that
>> DLEs are repeated.
>>
>
> Do you not have enough holding disk space to hold the entire dump set? You
> didn't say how big your tapes were, or how much data was getting dumped. If
> you have enough holding space, you can run the amdump, let it spool to
> holding disk, then just amflush until you've got everything to tape.
>
> The other option is that amdump can take hostnames as options, so you could
> run the job with a select number of hosts over a few day period. 


No.  I don't have enough disk space to hold the entire dump at once.
Sorry for not mentioning it.  There's enough space for the daily
amdump, but not for a complete archival dump.

Adding the hostname (and disk for the Windows machines) should solve
my problem.  Thanks for the suggestion.

-- 
Kevin Dalley
[EMAIL PROTECTED]


Re: Archival backup problems

2005-06-21 Thread Matt Hyclak
On Tue, Jun 21, 2005 at 01:26:49PM -0700, Kevin Dalley enlightened us:
> Thanks for the suggestion.  Setting record to no seems like a good idea.
> 
> However, dumpcycle set to 0 doesn't work for me.  amdump tries to run
> a full backup each day.  I only use 1 or 2 tapes per amdump, and I
> need 6-10 tapes for a complete archival backup.  So I need a few days
> to complete an archival backup.  Setting dumpcycle to 0 seems force
> the full backups to start all over again each day, which means that
> DLEs are repeated.
>

Do you not have enough holding disk space to hold the entire dump set? You
didn't say how big your tapes were, or how much data was getting dumped. If
you have enough holding space, you can run the amdump, let it spool to
holding disk, then just amflush until you've got everything to tape.

The other option is that amdump can take hostnames as options, so you could
run the job with a select number of hosts over a few day period. 

Matt

-- 
Matt Hyclak
Department of Mathematics 
Department of Social Work
Ohio University
(740) 593-1263


pgp8ua9Fu78wl.pgp
Description: PGP signature


Re: Archival backup problems

2005-06-21 Thread Kevin Dalley
Thanks for the suggestion.  Setting record to no seems like a good idea.

However, dumpcycle set to 0 doesn't work for me.  amdump tries to run
a full backup each day.  I only use 1 or 2 tapes per amdump, and I
need 6-10 tapes for a complete archival backup.  So I need a few days
to complete an archival backup.  Setting dumpcycle to 0 seems force
the full backups to start all over again each day, which means that
DLEs are repeated.


Matt Hyclak <[EMAIL PROTECTED]> writes:

> On Mon, Jun 20, 2005 at 03:50:41PM -0700, Kevin Dalley enlightened us:
>> I'm trying to run an archival backup which I run every 6 months or
>> so.  I hope to have a full backup of everything every 6 months.  This
>> is in addition to my standard backup, which includes incrementals, run
>> almost every week day.
>> 
>> It isn't working as I expect.
>> 
>> My daily amanda backup, however, is running fine.
>> 
>> Here are the parts of the amanda.conf which I think are relevant.  I'm
>> skipping the other 400 lines for now, though I will to post more lines
>> if someone needs more information.
>> 
>> I'm using amanda 2.4.5b1, with a couple of small changes I made.
>> 
>> The symptoms are that the same DLEs are backed up repeatedly, not
>> finishing the entire list of DLEs.  The DLEs aren't exactly the same
>> each day, but large DLEs are frequently backed up.  I use a different
>> config for the archival backups than I do for the standard backups.
>> 
>> 
>> dumpcycle 178 days
>> tapecycle 2000# number of tapes in rotation.
>> runtapes 2  # number of tapes used
>>
>
> You probably want dumpcycle 0 here.
>
>> define dumptype default {
>>  dumpcycle 100
>>  maxcycle 0
>>  skip-incr yes
>>  strategy noinc
>>  record yes
>> }
>
> I'm not familiar with maxcycle. Again dumpcycle should probably be 0 to
> force a full each time you run it. That said, you can probably leave out the
> skip-incr and strategy noinc. 
>
> Setting record to yes will mess with your regular daily cycle. Just make
> sure that's what you want.
>
> If I had to guess what's happening, is that it is selecting certain DLEs to
> be level 0's and they get backed up. The others are incrementals, so
> skip-incr is telling amanda ignore them. IIRC, strategy noinc is not
> implemented yet. That's why you're only seeing certain DLEs get backed up.
> Someone correct me if I'm wrong.
>
> For reference, my monthly archive is defined as:
>
> dumpcycle 0
> runspercycle 1
> tapecycle 13 tapes (I'm on tape 8 currently, I'll increase this as I get more)
> runtapes 1
>
> define dumptype global {
>   index yes
>   record no  
> }
>
> Matt
>
> -- 
> Matt Hyclak
> Department of Mathematics 
> Department of Social Work
> Ohio University
> (740) 593-1263

-- 
Kevin Dalley
[EMAIL PROTECTED]


Re: Archival backup problems

2005-06-20 Thread Matt Hyclak
On Mon, Jun 20, 2005 at 03:50:41PM -0700, Kevin Dalley enlightened us:
> I'm trying to run an archival backup which I run every 6 months or
> so.  I hope to have a full backup of everything every 6 months.  This
> is in addition to my standard backup, which includes incrementals, run
> almost every week day.
> 
> It isn't working as I expect.
> 
> My daily amanda backup, however, is running fine.
> 
> Here are the parts of the amanda.conf which I think are relevant.  I'm
> skipping the other 400 lines for now, though I will to post more lines
> if someone needs more information.
> 
> I'm using amanda 2.4.5b1, with a couple of small changes I made.
> 
> The symptoms are that the same DLEs are backed up repeatedly, not
> finishing the entire list of DLEs.  The DLEs aren't exactly the same
> each day, but large DLEs are frequently backed up.  I use a different
> config for the archival backups than I do for the standard backups.
> 
> 
> dumpcycle 178 days
> tapecycle 2000# number of tapes in rotation.
> runtapes 2  # number of tapes used
>

You probably want dumpcycle 0 here.

> define dumptype default {
>   dumpcycle 100
>   maxcycle 0
>   skip-incr yes
>   strategy noinc
>   record yes
> }

I'm not familiar with maxcycle. Again dumpcycle should probably be 0 to
force a full each time you run it. That said, you can probably leave out the
skip-incr and strategy noinc. 

Setting record to yes will mess with your regular daily cycle. Just make
sure that's what you want.

If I had to guess what's happening, is that it is selecting certain DLEs to
be level 0's and they get backed up. The others are incrementals, so
skip-incr is telling amanda ignore them. IIRC, strategy noinc is not
implemented yet. That's why you're only seeing certain DLEs get backed up.
Someone correct me if I'm wrong.

For reference, my monthly archive is defined as:

dumpcycle 0
runspercycle 1
tapecycle 13 tapes (I'm on tape 8 currently, I'll increase this as I get more)
runtapes 1

define dumptype global {
  index yes
  record no  
}

Matt

-- 
Matt Hyclak
Department of Mathematics 
Department of Social Work
Ohio University
(740) 593-1263


pgp5aQpeI9QXQ.pgp
Description: PGP signature