Front page

Order of clients/generations

f79319b63b894e89b68eb661affc0cf4
VIRUS MOLASSES BEDLAMP

From: Martin Hoßbach <mhossbach@gmail.com>
Date: Wed, 5 Jul 2017 12:11:57 +0200

   I'm currently "importing" data from one backup system into an obnam
   repository, using (among others) these two command line options:
   > --pretend-time="$TIMESTAMP"
   > --client "$CLIENT"
   
   Now, assuming I would do this client by client, and then generation by
   generation, it will happen that older generations from one client will
   go into obnam's repository _after_ newer generations from another
   client.
   
   Example -- this would be the order of obnam runs:
   * client A, date: 2017-01-01, midnight
   * client A, date: 2017-02-01, midnight
   * client B, date: 2017-01-01, midnight
   * client B, date: 2017-02-01, midnight
   
   I don't fully understand how the repository works, but I'm wondering
   if this could cause any problems? Could it cause confusions in obnam's
   "forget" code?
   
   Thanks,
   Martin
   
   _______________________________________________
   obnam-support mailing list
   obnam-support@obnam.org
   http://listmaster.pepperfish.net/cgi-bin/mailman/listinfo/obnam-support-obnam.org
From: Lars Wirzenius <liw@liw.fi>
Date: Wed, 5 Jul 2017 18:15:40 +0300

   On Wed, Jul 05, 2017 at 12:11:57PM +0200, Martin Hoßbach wrote:
   > I'm currently "importing" data from one backup system into an obnam
   > repository, using (among others) these two command line options:
   > > --pretend-time="$TIMESTAMP"
   > > --client "$CLIENT"
   > 
   > Now, assuming I would do this client by client, and then generation by
   > generation, it will happen that older generations from one client will
   > go into obnam's repository _after_ newer generations from another
   > client.
   > 
   > Example -- this would be the order of obnam runs:
   > * client A, date: 2017-01-01, midnight
   > * client A, date: 2017-02-01, midnight
   > * client B, date: 2017-01-01, midnight
   > * client B, date: 2017-02-01, midnight
   > 
   > I don't fully understand how the repository works, but I'm wondering
   > if this could cause any problems? Could it cause confusions in obnam's
   > "forget" code?
   
   First, genarations are per client, there's no ordering between
   generations between clients. As fars as "forget", you can try that
   wouth the --pretend option.
From: Martin Hoßbach <mhossbach@gmail.com>
Date: Wed, 5 Jul 2017 19:12:48 +0200

   That I already did and it seems to work. However: Since clients share
   chunks, there must be some mechanism for a chunk to know how many
   times it is used. Just want to make sure that part of the code can't
   be confused by using my approach.
   
   On Wed, Jul 5, 2017 at 5:15 PM, Lars Wirzenius <liw@liw.fi> wrote:
   > On Wed, Jul 05, 2017 at 12:11:57PM +0200, Martin Hoßbach wrote:
   >> I'm currently "importing" data from one backup system into an obnam
   >> repository, using (among others) these two command line options:
   >> > --pretend-time="$TIMESTAMP"
   >> > --client "$CLIENT"
   >>
   >> Now, assuming I would do this client by client, and then generation by
   >> generation, it will happen that older generations from one client will
   >> go into obnam's repository _after_ newer generations from another
   >> client.
   >>
   >> Example -- this would be the order of obnam runs:
   >> * client A, date: 2017-01-01, midnight
   >> * client A, date: 2017-02-01, midnight
   >> * client B, date: 2017-01-01, midnight
   >> * client B, date: 2017-02-01, midnight
   >>
   >> I don't fully understand how the repository works, but I'm wondering
   >> if this could cause any problems? Could it cause confusions in obnam's
   >> "forget" code?
   >
   > First, genarations are per client, there's no ordering between
   > generations between clients. As fars as "forget", you can try that
   > wouth the --pretend option.
   >
   > --
   > I want to build worthwhile things that might last. --joeyh
From: Lars Wirzenius <liw@liw.fi>
Date: Wed, 5 Jul 2017 21:25:07 +0300

   On Wed, Jul 05, 2017 at 07:12:48PM +0200, Martin Hoßbach wrote:
   > That I already did and it seems to work. However: Since clients share
   > chunks, there must be some mechanism for a chunk to know how many
   > times it is used. Just want to make sure that part of the code can't
   > be confused by using my approach.
   
   Unless there's a particularly horrible bug in Obnam, you're OK.