A simple wrapper script to keep retrying until a success/fail condition.

I originally wrote this as a simple wrapper script that monitored the return code from rsync and retried the sync if it was non-zero (i.e. failed to complete). This is useful if you have a very dodgy link which occasionally goes down causing the current rsync session to bail-out.

I have since decided it is much more useful as a general purpose retry wrapper so I've made it a lot more generic. The original behaviour can be run by:

retry.py -v -- rsync user@host:path/ .

Using -n and -c you can run something a number of times and collect a benchmark for fast something runs with what pattern of error codes. For example:

retry.py -n 5 -c -- ./random.sh

Gives:

Results summary:
0: 2 times (40.00%), avg time 0.019361 (0.000018 deviation)
3: 1 times (20.00%), avg time 0.025896 (0.000000 deviation)
4: 2 times (40.00%), avg time 0.011978 (0.000092 deviation)

I've added the -c/--count option for use in testing situations. If you have a program that fails occasionally you can run -n iterations and get a summary of the return codes at the end of your run.

Ideally your test program should return different error codes for different failure cases.