[Duplicity-talk] FtpBackend: extended timeout/retry on exceptions; abstracting error_wrap()

[dAniel hAhler]

Hi,

I've implemented a retry logic for FtpBackend and afterwards realized, that 
it's already present in CVS.

From the patch tracker item:
> I've changed the FTP backend to re-connect and retry the command on failure
> (exception), based on 0.4.2.
>
> Before submitting it, I wanted to merge it with CVS and there was already
> some similar logic.
>
> However, this patch retries on all exceptions, not just ftplib.error_temp.
>
> It also uses a bigger sleep interval with each retry.
>
> This should help a lot with having it running unattended in a cron job.
>
> Comments and commit welcome, of course.. :)
>
> I'll also write a mail to duplicity-talk, because I think the
> ftpBackend::error_wrap() method should be used the same way with all
> backends and therefor abstracted out of ftpBackend into Backend.


What do you think about abstracting FtpBackend::error_wrap() into Backend?

It seems to make a lot of sense, because it's all about backups and we really 
do not want to have it fail, because of some (temporary) error.

Luckily, the methods of the backends all seem to be atomic (especially put()). 

Thanks for a great tool!

pgpVZawye4alC.pgp
Description: PGP signature