Dear Harald,
Sorry for the delay, the UltraSite cabinet has some massive
backplane/interconnect issues, I tried to sort those out in the last
couple days (with not much success).
I had some time to further investigate the Nokia UltraSite situation,
and here are the outstanding issues I faced:
1. 2-3 seconds after BTS_RESET the OML LAPD link is reestablished
(coming from the BTS itself), and the signalling logic tries to
bootstrap OML (sometimes RSL as well) during the BTS performing the
reset. This tries do fail of course. Not sure if this causes any
practical error, although it does not look nice. Maybe adding some
logic state checks can help?
2. Sometimes on the OML LAPD link (and only on that one) I can see
these error messages:
DLLAPD lapd_core.c:1315 (0:1-T1-S62) S frame response with F=1 error
DLLAPD lapd_core.c:421 (0:1-T1-S62) sending MDL-ERROR-IND cause 6 from
state LAPD_STATE_MF_EST
Sometimes they arrive in batches, sometimes only a few. FYI the BTS
and the TRE both runs the latest SW ever released for UltraSite
(CX8.2).
3. Moving the RSL bootstrap to after "NOKIA_BTS_CONF_COMPL" received
seems to be working fine, this way we don't need to add another timer,
and Osmo-BSC can wait long enough to make sure all the TRXes loaded
the SW, configured and waiting for LAPD. In case more than one TRX is
used, the TRXes reach the "waiting LAPD" state in different order.
Did you had time to maybe try with InSite? I dont think my patch can
brake it, but I remember InSite was an odd ball within the Site
family.
I also started to re-introduce RF/BB hopping control for Nokia, if you
- or anyone -
can review it that would be lovely:
https://github.com/dchard/osmo-bsc/commit/41f6cd4723961700c5e5eceab4c92e7ce…
I was not able to try it out yet, as due to my backplane issues with
the cabinet, only TRX1 and TRX3 and TRX4 slots are working. And for
some reason if I try to use TRX3 and 4, the RSL bootstrap fails. I
wonder if the Nokia code has some limitation if the first TRX is not
actually the first? I tried to configure Osmo-BSC to start with "trx
2" but it does not want to start anything other than "trx 0". Of
course the question is what OsmoBSC sends during OML bootstrap. As
much as I can see, the Nokia code does not take into account if the
first TRX is not the first physically in the unit, but I am not a 100%
sure yet. Some help or hint here would be very welcome.
This is where I am now. The HW issues with the unit makes it pretty
time consuming to focus on the SW side, for now.
If you or anyone has any ideas or comments, please let me know.
Regards,
Csaba