Also I changed derby location to the relative Maven build dirs:
derby.jdbc.url = jdbc:derby:./target/derby/testdb;create=true
So there shouldn't be a need to access/manage /tmp. Now waiting for
some other projects' stuck builds to die so that the Cayenne-trunk can
start building on Ubuntu machines.
Andrus
On Jan 22, 2010, at 4:56 PM, Andrus Adamchik wrote:
FYI - I re-enabled parallel builds for speed, as clearly the Derby
problem wasn't fixed by sequential builds.
On Jan 21, 2010, at 10:32 AM, Andrus Adamchik wrote:
On Jan 21, 2010, at 10:26 AM, Aristedes Maniatis wrote:
I just turned on "Run each configuration sequentially". Before it
was running the builds simultaneously (and possibly spread over
different nodes). This will be much slower now, but if both builds
are trying to git the same db..
I observed a number of builds, and I think builds on the same
machine are never run in parallel. So parallel builds shouldn't be
a problem unless /tmp is shared between vesta and minerva. If tmp
is indeed shared, we can still ensure that separate dbs are
generated by editing copies of ~/.cayenne/connection.properties on
both build servers, changing this line:
derby.jdbc.url = jdbc:derby:/tmp/....;create=true
Andrus
