rtpengine not removing sessions despite no media

[Chris Uehlinger]

I'm running into an issue where my rtpengine instance is running out of ports. I allocate 23000-60000 and am testing with about 2000-5000 concurrent calls. I just did a test where I ran out of ports, and then tore down everything except the two rtpengine proxies (one of which proxies traffic to the other). I then ran `rtpengine-ctl list sessions all`, and then got the details for one of the sessions. There was no traffic for several minutes and no NG-Control traffic, but the session was never deleted.

Here's what the session looked like:


callid: 77...@172.31.2.65
deletionmark: no
created: 1726684043
proxy: 3.142.166.180:42554
tos: 184
last_signal: 1726684044
redis_keyspace: 0
last redis update: 0
foreign: no
recording: no

--- Tag '31SIPpTag0077', type: FROM_TAG, label '', branch '', callduration 1870.601025
---     subscribed to media with monologue tag '77' (index: 1)
---     subscription of media with monologue tag '77' (index: 1)
------ Media #1 (audio over RTP/AVP) using G729/8000
-------- Port     172.31.1.74:34752 <>     3.22.70.210:2000 , SSRC 9aba9ed, 3169 p, 101408 b, 0 e, 1726684048 uts 1726684107 kts
-------- Port     172.31.1.74:34753 <>     3.22.70.210:2001  (RTCP), SSRC 0, 0 p, 0 b, 0 e, 1726684043 uts 0 kts
--- Tag '77', type: TO_TAG, label '', branch '', callduration 1869.677219
---     subscribed to media with monologue tag '31SIPpTag0077' (index: 1)
---     subscription of media with monologue tag '31SIPpTag0077' (index: 1)
------ Media #1 (audio over RTP/AVP) using G729/8000
-------- Port     172.31.1.74:24188 <>  18.117.170.214:40644, SSRC 9aba9ed, 3162 p, 101184 b, 0 e, 1726684048 uts 1726684107 kts
-------- Port     172.31.1.74:24189 <>  18.117.170.214:40645 (RTCP), SSRC 0, 0 p, 0 b, 0 e, 1726684044 uts 0 kts

Here's my rtpengine.conf:

[rtpengine]

### Table 0 is for kernel space forwarding. You will often use this in production
table = 0

### for userspace forwarding only:
# table = -1

### separate multiple interfaces with semicolons:
interface = public/LOCAL_IP_ADDRESS!PUBLIC_IP_ADDRESS

listen-http = 0.0.0.0:22221
listen-ng = 22222
listen-cli = 22223

timeout = 1
silent-timeout = 1
offer-timeout = 10
tos = 184
delete-delay = 0

port-min = 23000
port-max = 60000
# max-sessions = 5000

# Run in the foreground, otherwise the container will exit
foreground = true

log-level = 6
log-stderr = true
log-facility = local4
log-facility-cdr = local6
log-facility-rtcp = local7
split-logs = true

mos = LQ

[Chris Uehlinger]

A couple other details: I'm running mr12.4.1.8 and there were about 4000 sessions persistently hanging around after I tore down the rest of the infrastructure.

[Richard Fuchs]

On 18/09/2024 15.28, Chris Uehlinger wrote:
> A couple other details: I'm running mr12.4.1.8 and there were about
> 4000 sessions persistently hanging around after I tore down the rest
> of the infrastructure.

Update to .9, or better yet the newest 12.5.

Cheers

[Chris Uehlinger]

I saw your comment on a closed GH issue referencing the "Good lord." commit (we've all been there), and updated to 12.5.1.7, that fixed this issue.

I'm having a new issue with rtpengine not recognizing call-IDs for answers, but I'll make a separate post for that when I have a writeup.