Re: 1.6.12: works; disk full handling

[On 30 Apr, @17:06, Boris wrote in "Re: [hdup-user] 1.6.12: works; ..."]
> 
> On Wednesday, April 30, 2003, at 07:47  AM, Miek Gieben wrote:
> 
> >This is what tar does on a full disk:
> >$ tar cvf trailer_final_1000_dl.mov.tar trailer_final_1000_dl.mov
> >trailer_final_1000_dl.mov
> >tar: trailer_final_1000_dl.mov.tar: Wrote only 2048 of 10240 bytes
> >tar: Error is not recoverable: exiting now
> >
> >>In case it matters, I was running a local, monthly backup with -V.
> >
> >it's a tar thing - and difficult to detect.
> 
> I don't working on a patch, but could use some guidance on the design.
> 
> Do you currently have anything in place that looks at the output of tar 
> and could detect that message?  Or could the message simply be passed 
> along to hdup's standard error?
> 
> How about exit status - I assume tar exits with a non-zero status when 
> it hits this unrecoverable error.  Could that be detected?

there are two exit status you can get from tar, well three actually,
0, 2 or 128. 0 is ok, 2 and 128 (i thought it was 128) mean something
bad has happened - not finding a file in a archive for instance yields
exit status of 2. Disk full also yields a exit status of 2.

> And what is the correct behavior is in this case?  It seems to require 
> some manual intervention no matter what, so perhaps leaving the partial 
> file there is fine as long as a loud warning is generated.  Possibly it 

i've implemented a statfs prior to the backup. With this hdup can check
if there is enough free space on the partition. This can be used by
saying "free = 5G" in the config file. If there is not 5G free on the
partition the backup is cancelled.

> should be renamed in such a way that it wouldn't be considered a 
> successful run by future invocations of hdup?  What do others think it 
> ought to do?

Something like that is maybe also a good idea.

grtz Miek