This is all good, and I think it underscores the point I was hoping to
make, which was that it would be very good to have a script in place
that sends out an email to derby-dev if performance for a given suite
regresses by some high-water mark (I was suggesting 150%).
As another example I remember Dan asking if a particular suite was
regressing due to some changes he was seeing being made that he was
concerned about.
David
Ole Solberg wrote:
Knut Anders Hatlen wrote:
"David W. Van Couvering" <[EMAIL PROTECTED]> writes:
I just noticed that in the latest tinderbox run that storetests had an
800% increase in the time it took to run compared to (some baseline?).
I think we should also get an alert email sent when a given suite is
more than 150% (or does someone have a better number) of the baseline.
Comments?
Sorry about that. I just happened to stress test a patch of mine on
the same machine as the tinderbox test was running on.
I have checked the history of 'storetests':
The increase occurs from 2005-12-01 - svn 351569/351543/"Fix for
DERBY-715" which added a new test:"... Added a test that causes 5
deadlocks..."
http://www.multinet.no/~solberg/public/Apache/TinderBox_Derby/UpdateInfo/351569.txt
http://issues.apache.org/jira/browse/DERBY-715
I guess the increase is explained by this.
A comment in the JIRA issue on an expected increase in execution time
for the test would be good though.
begin:vcard
fn:David W Van Couvering
n:Van Couvering;David W
org:Sun Microsystems, Inc.;Database Technology Group
email;internet:[EMAIL PROTECTED]
title:Senior Staff Software Engineer
tel;work:510-550-6819
tel;cell:510-684-7281
x-mozilla-html:TRUE
version:2.1
end:vcard
