Changes in trunk
dormando
I'm filtering bugfixes into trunk... They touch a handful of things, so be
wary if you're upgrading.
- The rare deadlock is now handled in an appropriate retry loop. I can
whittle at making them less likely, but it's very difficult (impossible?)
to fix all of them, so MogileFS should be tolerant of that.
This was painful, since I had to change the way the MySQL driver was
handling errors... Room for bugs here.
- Fsck status summaries are now more realistic, but still probably off.
- Fsck no longer shovels hundreds of millions of rows into file_to_queue
if you have hundreds of millions of files. It'll stop at 20k (by default)
I'm going through my inbox looking for bugs n' stuff. I want to hit up a
few poor error handling cases (like 'REQUEST FAILURE' for mogadm, and some
similar inside the monitor process), then cut another bugfix release.
If you have a particular bug you're troubling with right now, please give
me a ping about it.
-Dormando