Attention is currently required from: pespin.
fixeria has posted comments on this change by fixeria. ( https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/42492?usp=email )
Change subject: library/PFCP_Emulation: add f_PFCPEM_set_recovery_timestamp() ......................................................................
Patch Set 1:
(1 comment)
Commit Message:
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/42492/comment/22e5c9ec_32db0... : PS1, Line 12: peer restart between test cases.
Ah ok so it's not in between tests, it's in the same test because you are spwarning a different ConnHdlr to do the Assoc Setup...
Not really. All ConnHdlr connect to the same instance of PFCPEM, so this is not a problem. The association remains alive regardless of the `ConnHdlr` lifetime. Furthermore, the same association remains alive and is re-used even across multiple testcases, because osmo-s1gw currently does not send any heartbeats (patches for that are in review) and thus does not detect that the simulated UPF has been restarted.
Why would you like to avoid the IUT not detecting the restart? After all we are resetting the state in our PFCP endpoint after every test...
The existing testcase init logic does so-called "lazy" association. First we query the S1GW's PFCP state via REST (previously we relied on StatsD gauge for that). If S1GW reports no association, then wait for the assoc setup request and do the association. If S1GW reports an existing association (`state == connected`), we simply do nothing a proceed with the actual testcase logic.
It's fine to re-use the existing association, at least I see nothing wrong with that. The only problem here is that the local RTS of PFCPEM changes across testcases, and this will trigger S1GW to reset the existing association as soon as the RTS mismatch is detected. This is why I am adding API for changing the local RTS in PFCPEM, and the following patch makes use of it.