Hi Harald,
I made some progress with the Nokia UltraSite and managed to fix its backplane, so at least I was able to continue with reintroducing hopping control with the first 2 TRX slots of the cabinet.
During this process I realized that apparently even multi-TRX support is broken, which was working on the old OpenBSC codebase in 2016. If only the first TRX is configured the BTS bootstraps fine, but when a second TRX is configured, the unit goes into this weird state: the first TRX loops through configurng and waiting for LAPD states and stays in configuring state, while the second TRX is stuck ad supervisory state where TRX test is possible (it does pass the test).
I used my old multi-TRX config file that actually worked with UltraSite (and MetroSite) where the second TRX only has TCH/F so nothing fancy, no hopping or anything like that.
I wonder if my modification of moving the RSL bootsrap to after CONF_COMPLETE somehow changed something, maybe the BTS requires the RSL LAPD links to be up during the CONF state? Or the new FSM lchan refactoring might have introduced this in 2018...
So apparently we have some more basic issues here then hopping :-)
Regards, Csaba
On Wed, Jun 07, 2023 at 11:34:35AM +0200, Sipos Csaba wrote:
Or the new FSM lchan refactoring might have introduced this in 2018...
When implementing those, I tried as much as possible to keep the exact same behavior as before the lchan_fsms were implemented. But some corners of the code were weird / wrong and changed.
That was a long time ago, and we have implemented very many features and fixes on the Abis since then -- dyn TS, VAMOS, ... --, obviously never tested against Nokia. So it's quite possible that we broke it by accident, many times over.
But it should be obscure details -- the TTCN3 tests show that we generally conform to specs...
~N
-
Neels Hofmeyr -
Sipos Csaba