Unsupported Trxid

[Jon Melin]

Hello,

I have a few ideas of what I can do to remedy this problem but I would first like to hear what others are doing. My error logs fill up with a lot of messages just explaining that a TrxID is unsupported. I have a couple message types I don’t use and don’t create routes for. Is there a setting to not error on this, or are you dealing with them in another way (EO config)?

Thank you,

Jon

Topic

[Kevin Crist]

You can put a kill proc in before it gets routed. Killing those certain message types.

[Jon Melin]

Hello,

Are you saying one of the included procs, or just code up something that looks for those other specific message types and kills them in a pre-proc?

Thank you,

Jon

[Kevin Crist]

We just code up something for the certain message types you want. I guess you could also route your unwanted mesasges to a bitbucket and use the generic hcimsgkill provided.

[Rehman Masood]

I would just set a default route at the end and route it back to the thread with hcitpsmsgkill.

[Elisha Gould]

A method we use is to add a _HCI_static_route_ set to “Generate” using the hcimsgkill proc.

This kills a copy of every message routed, so it adds a little extra overhead, but it stops the routing error.

The best option would either be to kill before the routing or to set a kill route that contains the routes not being used.

[Charlie Bursell]

Elisha:  You must have an old version if you use the Generate route.   🙂

If you are going to run it through hcitpsmsgkill a raw route would suffice

I do agree it would be somewhat more ergonomic to put in a filter proc and just KILL the ones that need it.  But, as you said, it adds a little overhead and if throughput is not a problem, go for it

The downside is if a vendor sends you a real outlaw message you would never know it since there would be no TRXID errors

[Paul Bishop]

We always make sure that we have _HCI_static_route_ setup on our threads with routing.  If any of the the destinations don’t need/want all messages, then we route the thread to itself with _HCI_static_route_ and hcitpmsgkill as a pre-proc.

Paul Bishop
Carle Foundation Hospital
Urbana, IL

[Jon Melin]

Great ideas. Thank you everyone. This is around the same lines as what I was thinking I’d setup. I just didn’t know if there was a universal way others were using.

[Keith McLeod]

In line with what Charlie stated, we typically define a route or wildcard route specific to those message types that we wish to kill and route it back to the source thread itself with a raw route and a hcitpsmsgkill proc.  This way if new trxids start to show up, you at least have an inventory of what is actually being sent…and then can determine if they would be useful for any receiving system.

It also allows you to keep track and force testing of any new messages before sending them on through.

[Author]

Replies