Cellular Network Infrastructure » cni-legacy » OpenBSC

The current status is:

• osmo-msc: Fully configurable, configuration may even be left out completely. In this case osmo-msc will automatically assume a default configuration.
• osmo-bsc: Configuration may not be left out. The user must set up a local address. The VTY still has some issues to be fixed. The goal here is to reach a state, where it works without configuring anything, just like it already is with osmo-msc

In order to move on with this properly, we must agree on a a default configuration for each component and on a default environment. The default environment should be the "Everything on one box" situation. Which means the ip-addresses will be all loopback addresses. We also must agree on pointcodes for each application. I would suggest to put this into a header file, but then there is the problem where to put it. libosmo-sccp sounds like a good place. Maybe having it just as a wiki-page is also enough.

• Do not mix up different addresses from different CS7 instahces (means, local and remote address must come from the same CS7 instance)
• Use only the msc address to determine the CS7 instance. Using the local address does not make too much sense especially when it is generated automatically from the selected CS7 instance.

We now prevent mixing addresses from different CS7 instances.

neels patches for this are on pmaier/aoip3

• We now have CS7 default values in osmo-bsc that are complementary to the ones used in osmo-msc. The user may leave out the CS7 configuration completely when all processes are used on the same machine.

• Fixed pmaier/aoip so that it compiles again with the current master libosmo-sccp.

• It does not make much sense to offer auto configuration for more than one MSC since it is a very rare and unlikely situation => Only allow one instance to be auto configured

Agreed with neels for the following default pointcodes:

MSC: 1.0.0
BSC: 3.0.0
SGSN: 2.0.0
HNBGW: 4.0.0

...would be more "realistic"

Yes, I think so too. We should use that set of pointcodes.

We still have a bug with osmo-bsc, according to neels the configuration files can not be read back because it prints a "(null)" into the config file during write. Should be simple to fix. Apart from that I think we made good progress with the VTY and the automatic configuration simplifies everything a lot.

The problems with the "(null)" strings are fixed. I think the VTY in terms of CS7/SS7 is pretty much done now. We will see how it performs in real life situations. Maybe there are some things we did not have on the radar yet, we will see. Lets keep this at 90% for now.

Sinde there is a bug in libosmo-sccp that seems to prevent the usage of pointcodes longer than 8 bit (see http://osmocom.org/issues/2441) We use the following default pointcodes now:

MSC A-Interface: 0.23.1
MSC Iu-Interface: 0.23.2
MSC A-Interface+Iu-Interface: 0.23.1
BSC: 0.23.3

There were no acute problems (other than a minor merge conflict in osmo-msc) with this in the past week. Lets consider this as resolved and add new tasks in case of problems/feature-requests