Re: Can a post-dle-backup script invalidate the backup?

More thinking on this:

The order of operation was a follow:
  
  sendbackup run pre-script
  application do the backup
  application save the state file if the backup is saved on server
  applcation send its state SUCCESS, PARTIAL, FAIL to server
  server log backup result
  sendbackup run post-script

The problem was the state file is written even if the server do not successfully write the backup on holding disk or tape.

I fixed it in trunk with the following:

  sendbackup run pre-script
  application do the backup
  application send its state SUCCESS, PARTIAL, FAIL to server
  application ask server if the backup is correctly saved (on holding disk or tape)
  server log backup result
  application save the state file if the backup is saved on server
  sendbackup run post-script

The application must be fixed to work with that new protocol

This have 2 problems ?
  - Failure to write the state file is not logged correctly
  - post-script cannot correctly fail backup

I propose to change the execution order to:

  sendbackup run pre-script
  application do the backup
  application ask server if the backup is correctly saved (on holding disk or tape)
  sendbackup run post-dle-script
  sendbackup send post-dle-script result to application
  application save the state file if the backup is saved on server and post-dle-script succeeded
  application send its state SUCCESS, PARTIAL, FAIL to server
  server log backup result

That way, the application will send a FAIL if the post-script fail, there is no reason to force a backup le
vel in the next run.

What do you think?

Jean-Louis

________________________________________
From: Chapman Flack <[email protected]>
Sent: December 21, 2017 3:45 PM
To: Jean-Louis Martineau; Amanda Hackers
Subject: Re: Can a post-dle-backup script invalidate the backup?

On 12/21/2017 03:27 PM, Jean-Louis Martineau wrote:

> The application save its state file (gnutar listdir) before it exit, and
> the following backup can be relative to that state file, failing a
> backup after that step could lead to unreliable following backup.

I would want the effect of invalidating a backup to include forcing
the next one to be at a level no greater than the invalid one,
so any saved state from the invalid one would be moot.

-Chap
This message is the property of CARBONITE, INC. and may contain confidential or privileged information.
If this message has been delivered to you by mistake, then do not copy or deliver this message to anyone.  Instead, destroy it and notify me by reply e-mail