[Spce-user] Are the redis counter reliable?

[Alex Lutay]

Dear Abel,

Can you please check the following content:
> redis-cli -n 4 keys \*

It should have call-id for the counters were not cleaned from Redis
(note: you should check then when you have no calls on system,
otherwise you will get all active calls too).

Then check the call-id(s) you have found on kamailio-proxy.log
and check kamailio restart nearby. We know one issue for PRO
(MT#15389 for references, already fixed in trunk, testing)
which appears in a moment of kamailio restart.
(You can detect restart using line: Thank you for flying kamailio)

Maybe CE users also affected by the issue.

If no restart happens nearby, we would to have call scenario shared,
to find the reason why counters were not removed from DB.

At the moment we know only one issue with counters on PRO/Carrier during
heartbeat switchover, maybe it exists on CE too during kamailio proxy
restart.

P.S. thank you for sharing NGCP version :-)

On 10/08/2015 03:59 AM, Abel Alejandro wrote:
> The ngcp-kamctl counter is the most reliable and always gives me the
> correct one, but the redis is way off. 
> 
> root at sip:~# /usr/sbin/ngcp-kamctl proxy fifo get_statistics
> :dialog::active_dialogs
> dialog:active_dialogs = 0
> root at sip:~# redis-cli -n 3 get total
> "15"
> root at sip:~# redis-cli -n 3 get outgoing
> "2"
> root at sip:~# redis-cli -n 3 get incoming
> "13"
> 
> root at sip:~# cat /etc/ngcp_version 
> mr4.0.1
> root at sip:~# 
> 
> Inline image 1

-- 
Alexander Lutay