For instance: http://testing.libreswan.org/results/testing/3.20-28-g41be757-master/klips-netkey-pluto-06/OUTPUT/west.console.diff I suspect this is low hanging fruit and I suspect it is a common problem:
I believe the following is happening and it should be deterministic: westrun.sh runs on west: - sets up a connection and leaves it open final.sh runs on east: - does ipsec look - shuts down pluto which shuts down the connections - ... final.sh runs on west: - does ipsec look; since the output expects the connection to still be present this fails to match (there could also be a race here since west:final.sh might run before west has finished shutting the connection down?) - shuts down pluto - ... (swantest would run "final.sh" on east and west in parallel making the behaviour non-deterministic). I can see several fixes: - add "ipsec look" to westrun.sh so it is performed before east shuts down - update west's expected output so it matches what is really happening when final.sh is run (but I wonder if that part can be dropped). Andrew _______________________________________________ Swan-dev mailing list [email protected] https://lists.libreswan.org/mailman/listinfo/swan-dev
