Re: [Gems-users] Odd EL_Timestamp behavior

["Jim Leek" <jrleek@xxxxxxxxx>]

On Wed, Mar 12, 2008 at 12:05 PM, Jayaram Bobba <bobba@xxxxxxxxxxx> wrote:

It is weird. Timestamp policy should not livelock... Can you get a more
detailed output and see if it is really livelock
or if the oldest transaction died and the other transactions abort
waiting for it?

   What kind of output would you like?  I was just dumping stats, continuing, then dumping stats again.  So, after letting barnes go for 20 hours, I got:

xact_size_dist: [binsize: 2 max: 68 count: 224 average: 7.95536 | standard deviation: 10.643 | 0 51 0 145 0 0 15 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 0 1 1 1 0 2 ]
xact_aborts:        16013

Of course, maybe it didn't live lock, it just froze for some other reason?  I don't know how to tell.

This is a el barnes that succeeded, so those numbers aren't all that big....
 xact_size_dist: [binsize: 2 max: 73 count: 2694 average: 9.30401 | standard deviation: 12.6054 | 0 387 85 1784 0 0 123 180 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 4 2 7 7 17 10 27 20 14 10 9 6 0 0 0 0 0 0 0 0 0 0 0 0 0 ]
xact_aborts:        87920

The pathology numbers were computed from trace files output by
XactVisualizer. However, I dont think the entire functionality
of the visualizer is included in the release. We didn't have the time to
clean it up and put it out.

Ok, I was just curious if I could get a better idea of what was going on.

Jim