Re: [rdiff-backup-users] Very very weird - rdiff working sometimes

[Andrew Ferguson]

On Apr 25, 2008, at 6:36 AM, address@hidden wrote:

2^nd Client – Two installs

                1.            Rdiff-backup for windows (1.0.4) full backup took 2 days, has been working for a couple of months no problems (scheduled with tasks)

                2.            Full Cywin install rdiff backup (1.0.5) – full backup took 3 days – now incremental fall over does not complete and then goes to regressing destination ( have followed instructions about deleting current mirror and moving mirror_metadata. Backup completes when run manually, does not complete with cron job?

 

3^rd Client -  – Full Cygwin install rdiff backup (1.0.5), Initial backup took 7 days, then same errors as 2^nd Client 2^nd Machine. We initially tried with the rdiff-backup for windows but it we need to be able to access machine via ssh so the full cygwin was installed.

 

Remote server running rdiff-backup 1.0.5

 

I cannot understand how it works for some and not for others.

 

Script used in majority of installs (obviously backup from location is different from client to client) the only thing that changes is the - - force is not used at the 1^st client

 

rdiff-backup --force --exclude-globbing-filelist backupexclude /cygdrive/c/Documents\ and\ Settings/ / <client_username>@<server_IP>::/home/officebackup

 

I am not too sure if the --force should be used all the time or only for the initial failures

Andy,

I think --force is your problem, and that cron is too dumb to avoid getting you into the situation. --force should only be necessary after certain failures. Without --force, rdiff-backup will still recover from incremental failures. It knows to regress the previous operation and start over. The --force is necessary, as I suspect you found out, when the *first* backup failed, and you don't want to delete and start over.

If rdiff-backup is run with --force, at least in the development version, it will allow a second instance of rdiff-backup to over-write an already running instance!

Since your link seems to be slow, or the source has a lot of data, I imagine that rdiff-backup doesn't complete sometimes in between the frequency with which it is being called by cron. It could be possible that two instances of rdiff-backup are dueling.

It would seem from your report that Windows tasks is smart enough to let a given task complete before starting it again.

If the above scenario is completely outside the realm of possibility, please post an tracebacks or error messages from rdiff-backup for further diagnosis.

best,

Andrew