Excessive checks of unmounted disk in reconstructor

[Caleb Tennis]

When a disk goes unmounted in the system, the logs fill up with these:

Sep 2 19:45:46 localhost object-reconstructor: 172.30.3.45:6003/d17/424 policy#1 frag#10 responded as unmounted

Which is an early check in the "_get_suffixes_to_sync" method.

A handful of disks makes the logs bloat dramatically. However, it seems like it would be more performant if we knew that a disk was unmounted that we didn't continue to check for subsequent jobs but just accepted it (at least for a period of time). This would speed up the jobs as we wouldn't need to continue querying the remote server.

[Bill Huber (wbhuber)]

I'm unable to recreate the issue at the moment. The log message from object-reconstructor seems fixed by other patches between Sept 2 and today. I'm marking this as incomplete as further notice.

[clayg (clay-gerrard)]

I can still seem 'em

object-6020: 127.0.0.1:6030/sdb3/267 policy#1 frag#1 responded as unmounted

I think my tree is current, and I'm not sure which change would have fixed it. Pretty sure I wouldn't have missed something that added the "caching 507's" suggestion...