2 very simple changes.
The first redirects stderr to stdout for the comparison tool so that c-i can see that the test is doing something. The second keeps the port# from exceeding 65535 on a busy machine.
2 very simple changes.
The first redirects stderr to stdout for the comparison tool so that c-i can see that the test is doing something. The second keeps the port# from exceeding 65535 on a busy machine.
Otherwise travis thinks the build has stalled.
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/p4612_6548cc9f9073bf6602e17f62017af5c0e2bc3457/ for binaries and test log. This test script verifies pulls every time they are updated. It, however, dies sometimes and fails to test properly. If you are waiting on a test, please check timestamps to verify that the test.log is moving at http://jenkins.bluematt.me/pull-tester/current/ Contact BlueMatt on freenode if something looks broken.
8 | @@ -9,7 +9,7 @@ mkdir -p "$DATADIR"/regtest 9 | touch "$DATADIR/regtest/debug.log" 10 | tail -q -n 1 -F "$DATADIR/regtest/debug.log" | grep -m 1 -q "Done loading" & 11 | WAITER=$! 12 | -PORT=`expr $BASHPID + 10000` 13 | +PORT=`expr 10000 + $BASHPID % 55536`
For a minute I was confused whether it is allowed to use port 65535, but apart from some remarks that 49152+ is 'ephermal port range', which is advisory, I could not find anything conclusive, and binding on port 65535 was no problem when I tried. But it may nevertheless make sense to use a safety margin here.
Sorry I missed your comment here before merge. I can slim down the range if you would prefer.
Nah. On a busy system, port collisions can always happen no matter what range is used (this just avoids colliding with another bitcoind instance), so I don't think toning down the range would help.