On Thu, Dec 05, 2013 at 09:13:59AM -0800, Glenn Tarbox, PhD wrote:
> obworker typically runs for a day or so before it experiences an
> error. Looking at the code, it appears there should be a way to
> restart without dumping the process... although that would be
> preferable to the thread-join state it currently hangs in.

Is the blockchain already downloaded in its entirety? I've had similar
errors (on a 64-bit Debian Wheezy) while still downloading, but not once
since then. It's been running stable for a little less than a week.

> There may be a reason network exception handling isn't enabled
> (leveldb state rollback makes it hard? Other state? Low priority?)
>
> I could take a deeper look but there might be a plan so I figured I'd
> ask first

From the low volume on this list I reckon there is no concrete plan, nor
enough developers for there to be a need for one; and Obelisk is in a
testing phase.

But I've only gotten interested about a month ago (as most people), so
don't trust me on this one (yet).