[Box Backup] manual snapshot problem

Torsten boxbackup at boxbackup.org
Tue Oct 21 10:57:49 BST 2008


Hi Chris,

Am Saturday 18 October 2008 12:32:48 schrieb Chris Wilson:
> Hi Torsten,
>
> On Sat, 18 Oct 2008, Torsten wrote:
> > You are right. I have errors in the backup runs. I still have to locate
> > them ;)
>
> Doesn't the backup output/system logs help?

Not really. It only shows an 

bbackupd[9649]: WARNING: Exception thrown: CommonException(OSFileError) at 
BackupClientFileAttributes.cpp(307)
bbackupd[9649]: ERROR: Exception caught (Common OSFileError (Error accessing a 
file. Check permissions.) 1/9), reset state and waiting to retry...

But i do not know what file has wrong permissions. ExtendedLogging did not 
help.

> > I do not know if this works for me. Does UpdateStoreInterval only
> > influence the backup interval?
>
> Yes, as far as I know.
>
> > I mean, is
> >  "Automatic Backup with UpdateStoreInterval=3600"
> > and
> >  "AutomaticBackup=no, UpdateStoreInterval=99999999999999 and manual
> > snapshot once per hour with crontab"
> > exactly the same?
>
> It's almost but not quite the same. There is some randomisation of times
> introduced into the AutomaticBackup mode which is supposed to prevent
> clients from falling into lock step with each other. Also, the latter does
> rely on external influence, whatever that is, to kick off a backup, and
> it's possible to do something much more sophisticated than just using cron
> to do that.

Right, but that is no problem for me.

> > But i think this configuration is not really intuitive. I mean it could
> > be difficult to describe how this works. Perhaps another
> > ConfigurationOption (e.g. ImmediatelyRestartSyncOneErrrors) would be
> > clearer?
>
> That could work, but I'm reluctant to introduce a new option until I've
> exhausted all options to make the default behaviour work for everyone.
>
> The idea of the current behaviour is that if you requested a backup, and
> it didn't complete successfully (e.g. the connection was interrupted),
> then it should keep retrying until it does. I can see that there might be
> a case for a retry limit here.
>
> It should already ignore errors like failures to read files, and only do
> this if something really bad like an exception causes the backup to fail
> early. Is this a case where your DSL provider is killing your connection?
> If so, then don't you want to try again immediately and keep trying until
> you've completed a backup?
>
> Is this happening in the middle of uploading a large file? Is it only a
> problem because Box Backup doesn't know how to resume partial uploads?

But i think my problem is no connection problem. My problem at the moment is 
that at the box backup server i have to pay for traffic. And so restarting 
backup run immediately after error generates lot of traffic.

Restarting Backup on errors after UpdateStoreInterval would be a solution for 
me. How complicated is it to implement this function?

nice day
  Torsten



More information about the Boxbackup mailing list