frr
frr copied to clipboard
FRRouting
topotests: test bfd when bgp is passive
Some versions between 10.0 and 10.3.1 had BFD not working if one of BGP peers was in "passive mode", i.e. instead of peers with addresses, a peer-group was defined to be matched by bgp listen statement. This problem went unnoticed because there was no test for bfd in such mode.
This commit modifies one of existing bfd tests to run bgp session "passively". It fails against 10.3, and passes against 10.3.1
While this is okay, can we have a separate topotest instead of replacing this one (non peer-group...)?
While this is okay, can we have a separate topotest instead of replacing this one (non peer-group...)?
I actually started by making a copy of the test in a separate subdir :grin:
What would be a good way to make a separate test that uses all the same code, and only differs in config files? It seems too ugly to just copy everything? What's your advice, @ton31337 ?
Create a new test_XXXX.py in the same directory. Have it setup slightly differently but all the same code put into a XXX_common.py ( Look at what was done in the route_scale test ).
Create a new test_XXXX.py in the same directory. Have it setup slightly differently but all the same code put into a XXX_common.py ( Look at what was done in the route_scale test ).
Did that. With the existing infrastructure, it is still more verbose than I would like, but I tried to make it as succinct as was practical :slightly_smiling_face:
Can this be merged now please, or are there other remarks? I believe that test failure is unrelated to this change. Thanks!