Hi,
I was wondering what precisely these stats mean:
core:drop_replies = 0 core:drop_requests = 3
Assuming one does not explicitly drop() requests anywhere in script, would a continuous monotonic increase in drop_requests be considered a problem? Or are there some normal circumstances and ordinary behaviours (i.e. auth challenges) that would register a request drop?
Thanks,
-- Alex
Hello,
iirc, should be the case of requests dropped from pre-script callbacks or from onsend_route ... perhaps a look in the code is the best.
Anyhow, should not be something to worry if there are no alarming log messages.
Cheers, Daniel
On 13/07/14 02:54, Alex Balashov wrote:
Hi,
I was wondering what precisely these stats mean:
core:drop_replies = 0 core:drop_requests = 3
Assuming one does not explicitly drop() requests anywhere in script, would a continuous monotonic increase in drop_requests be considered a problem? Or are there some normal circumstances and ordinary behaviours (i.e. auth challenges) that would register a request drop?
Thanks,
-- Alex
My concern is that this figure reflects requests which are not handled at all (prior to the expiration of some timer and a retransmission) because there are no worker threads able to get to them. Is it possible?
On 18/07/14 20:48, Alex Balashov wrote:
My concern is that this figure reflects requests which are not handled at all (prior to the expiration of some timer and a retransmission) because there are no worker threads able to get to them. Is it possible?
No, if there is no worker to handle the packet, then it is not known that it was dropped (in case it was from network buffer) or if it is was already in transactions hash table, there has to be a kamailio process getting it out of there.
A reason that could be for dropping is the case of invalid sip messages -- initial parsing fails and the message doesn't get to routing script.
Cheers, Daniel
-
Alex Balashov -
Daniel-Constantin Mierla