lockfile : wait instead of aborting

[unterkomplex]

Hi,

Feature request - currently btrbk abort if the lockfile is present

Change this (optional or by default) to wait until the lockfile has been removed, then continue

This would be useful for example when connecting multiple backup disks and copying to all of them in parallel
btrbk archive --override=lockfile=/run/lock/btrbk.lock --override=lockfile_present=wait source target

[schildbach]

Note that the current behaviour of "aborting in case a lock is already acquired" also protects us from resource exhaustion if a backup hangs or is just very slow running for some reason. In a sense, the current behaviour is self-adjusting should the process take longer than the time slot that is allocated to it.

So while I support this idea in general, I think some more advanced form of protection needs to be taken into account – e.g. limiting the number of waiting processes waiting for the lock, or wait time-outs.

[schildbach]

The example you're giving could also be implemented by this idea: #639
and better parallelization in general by building a dependency tree and then running operations concurrently.