prevents workers to stale when redis server went away

[jesusch]

fix for #180

unfortunatelly I have no clue, why the first access on redis()->ping() does not throw the exception anyhow this will prevent worker from getting useless after a network failure or redis restart

[danhunsaker]

Throwing a CredisException instead of duplicating the error handling code would probably be a good idea.

That said, it's interesting that Credis itself doesn't handle this... The reason we don't is it's supposed to do it for us.

[jesusch]

I've updated the pull according to your comment

[danhunsaker]

Looks good to me.

I still find it odd that Credis doesn't handle that on its own, though. I wonder if it's handled correctly in a more recent version of the library...

[jesusch]

any chance that this pull requests gets merged?

[jesusch]

I had to update the commit due to missing constructor message on CredisException

[danhunsaker]

:+1:

[chrisboulton]

I think in theory this should also be covered by #229, which was just merged down - essentially, the next operation against Redis on an instance that's gone away will cause the worker to throw an exception (Resque_RedisException)