[Box Backup] Re: [Box Backup-dev] Re:BoxReport.pl (was ChrisMerge_1828)
Matt Brown
boxbackup@fluffy.co.uk
Thu, 18 Oct 2007 21:52:29 +0100
Hi Chris,
>> The script can be run by hand with or without command line
>> parameters (will default to the last found backup in the log
>> file). Script does require ExtendedLogging = yes to get the data
>> it needs.
>
> You might want to look into using LogAllFileAccess messages instead
> of ExtendedLogging. ExtendedLogging might be slower (it's on the
> network path) and LogAllFileAccess should include full paths in
> each message. It's also more readable, IMHO. Hopefully the output
> is not difficult to parse.
>
>> -- Chris I might try altering the code for the backup-client to
>> call boxreport.pl if it finds ReportScript = /usr/local/bin/
>> boxreport.pl (or similar) in bbackupd.conf (if I can work out how
>> to do it :-) that way I can get the the report to generate and
>> mail right after the backup has completed rather than having to
>> wait for cron etc.
>
> There should be no need for that, now that we have backup-start and
> backup-finish hooks for NotifyScript. When NotifyScript is called
> for backup-start, it could record the current time. When called for
> backup-finish, it could generate a report from the backup-start
> time to the current time.
Following on from this, is it possible to send an interrupt signal to
bbackupd ? i.e something like bbackupctl -q stop or bbackupctl -q
interrupt ?
The reasoning behind it is we only allow clients to backup between a
certain window each day, more for there benefit so it does not
saturate their line and cause us to get "My internet connection is
really slow etc" type calls. At the mo I simply run a cron at two
certain times a day, and issue a boxbackup-client stop which in turn
just kills the process and isnt very clean.... its not very often a
backup takes a huge amount of time, until someone decides to offload
their entire collection of MP3, Docs, Pron etc somewhere on the mount
point :-)
Then NotifySysAdmin could get a handle like backup-stopped or backup-
interrupted ? The report can then take this action into account and
also means we can keep the daemon running without having to keep
starting and stopping each day.
Your thoughts ?
Matt