Paul, The quick test with the offending patch removed looked good. It just found the already reported memory leak. I've started a more comprehensive test on whole branch -- will take a~2.5 hours to run.
I'm also rerunning the step-wise basic (4min) regression through every other commit to find the origin of the memory leak. -- just 109 (out of 160) remaining! Lou On 7/13/2016 6:22 AM, Paul Jakma wrote: > Hi Lou, > > Thanks. Donald had another patch in that series, 0b697fa9281a62 / "ib: > Add ability to use poll() instead of select", but had indicated that was > kind of preliminary still and that he wanted to do further work on it. > > If you drop the below, do you find further probs? > > regards, > > Paul > > On Wed, 13 Jul 2016, Lou Berger wrote: > >> This is patch that's breaking bgpd in 8/proposed/ff >> >> Author: Donald Sharp <[email protected]> >> Date: Fri Mar 4 15:28:56 2016 -0500 >> >> lib: Refactor read/write functionality >> >> Both the read and write functions used the same code >> slightly modified for reading and writing. Combine this >> code together. >> >> Signed-off-by: Donald Sharp <[email protected]> >> >> Edited-by: Paul Jakma <[email protected]> to retain the >> external library symbols, for ease of merging. >> >> >> On 7/12/2016 7:55 PM, Lou Berger wrote: >>> Just an update: we've hooked our regression system into the github >>> mirror and are now running minimal regression tests on bgpd. The tests >>> start at a commit and move to the head of the branch, commit by commit >>> -- pretty simple approach. >>> >>> Each run does a compile, basic adjacency checks, some unicast and VRF >>> route distributions and checks the results against a reference (known >>> good run). Each run takes about 4 minutes and there are about 160 >>> commits in /8/proposed/ff - currently has about 150 to go. >>> >>> I'll provide an update once we have interesting results. >>> >>> Lou >>> >>> PS I rebased the following commits to be 1st in order to get the >>> regression environment running: >>> >>> Author: Lou Berger <[email protected]> >>> Date: Tue May 17 07:10:37 2016 -0400 >>> >>> bgpd: Add flag to not change e{u,g}id on startup and run as >>> unprivileged user >>> >>> * bgp_main.c: add -S / --skip_runas flag to not change effective >>> user/group >>> on start up. Enables bgpd to be run by unprivileged user. >>> >>> Author: Lou Berger <[email protected]> >>> Date: Tue May 17 07:10:36 2016 -0400 >>> >>> bgp: add "debug bgp allow-martians" next hops and related code/commands >>> >>> Author: Lou Berger <[email protected]> >>> Date: Tue May 17 12:19:51 2016 -0400 >>> >>> lib: change command logging to be off by default >>> >>> * lib/vty.c: add 'log_command' to enable logging of vty commands >>> executed. >>> Default command logging to off. >>> >>> >>> >>> On 7/11/2016 5:44 AM, Lou Berger wrote: >>>>>> Any luck pinning down what commits are causing which issues for you? >>>>>> >>>> Not yet. Thinking I'll try to wire into our regression system in some >>>> way... >>>> >> >> >> _______________________________________________ >> Quagga-dev mailing list >> [email protected] >> https://lists.quagga.net/mailman/listinfo/quagga-dev >> _______________________________________________ Quagga-dev mailing list [email protected] https://lists.quagga.net/mailman/listinfo/quagga-dev
