Hi Carsten,

first of all, it was a mistake in my previous email - the id is not randomly inited, but with 0. I made a confusion with hash labels in TM....:)

so, we do agree that the the id (as populated in receive.c) cannot overlap...the only possibility may be TM related. In failure_route , a fake request is build to be passed to the script (as no real request was received, being triggered by a reply). This fake request has " id = original_req_id -1 ".....so, in failure route you may have a request with an id already used......

can you check if the errors are failure route related??

regards,
bogdan

Carsten Bock wrote:

Hi Bogdan,

I did not restart the proxy and even if we did, this would not have lead into an error, because then we would do a database lookup in every case (unless the ID would be 0). I know, it should be incremented for every received message and it should not lead into this error, but in our case it did not work properly :-( I am investigating on this issue and keep you'll informed if i find something.


CU in Berlin,
Carsten


Bogdan-Andrei Iancu schrieb:

Hi Carsten,

have you restarted your proxy between the two requests with same IP?
the msg id is initialized in each process with a random value and incremented for each received message, so it will loop after ~4 billions messages :).

regards,
bogdan

Carsten Bock wrote:

Hi everybody,

I have just seen a strange phenomenon with "msg->id" field, which, according to the source should be unique per thread.

From parser/msg_parser.h, line 161:
[...]
struct sip_msg {
   unsigned int id;               /* message id, unique/process*/
[...]


We use this field for minimzing Database-Lookups so we do some lookups just once per request. But now we found some very few examples (e. g. maybe 2-3 times per 1000 Requests), were this id was the same for different requests per thread and this caused some trouble on our proxy. I did see, the call-id of the request was different, but the id was the same. The calls were generated by an Asterisk (SIP) calling another Asterisk (E1) via an Cisco Mediagateway.
Has anyone else experienced this?

Thanks,

Carsten




_______________________________________________
Devel mailing list
[email protected]
http://openser.org/cgi-bin/mailman/listinfo/devel






_______________________________________________
Devel mailing list
[email protected]
http://openser.org/cgi-bin/mailman/listinfo/devel

Reply via email to