Heplify blocks port 8090 for sipwise sems-b2b

[ciscospirit]

Hello,

we found out that there is somewhere a problem if we doing a switchover from our active node sp2 to our passive node sp1.

on the passive node (sp1), heplify is running and sems-b2b not. as soon we make this node active, sems-b2b fails to start, as heplify is already using port 8090 on tcp6 to LISTEN. but 8090 is manadory for sems-b2b to work correctly and to start correctly.

how can we prevent this? Also why it is listen to tcp6? maybe it is a startup problem, that heplify is not allowed to start before sems is started?

any idea how to fix this?

# ngcp-parallel-ssh all "ps axuf | grep heplify" | grep -v grep
[1] 15:28:56 [SUCCESS] sp1:22
root     3409461  0.1  0.2 2416760 78424 ?       Ssl  15:27   0:00 /etc/ngcp-config/xxx/homer/heplify/heplify -hs homer.xxx.xx:9060 -m SIP -hn sppro -hi 2002 -p /var/log/homer/

[2] 15:28:56 [SUCCESS] sp2:22
root     1350697  1.1  0.9 2760416 326792 ?      Ssl  Feb29 432:39 /etc/ngcp-config/xxx/homer/heplify/heplify -hs homer.xxx.xx:9060 -m SIP -hn sppro -hi 2002 -p /var/log/homer/

# ngcp-parallel-ssh all "netstat -ploan | grep 8090"
[1] 15:29:23 [SUCCESS] sp1:22
tcp6       0      0 :::8090                 :::*                    LISTEN      3409461/heplify      off (0.00/0/0)

[2] 15:29:23 [SUCCESS] sp2:22
tcp        0      0 127.0.0.1:8090          0.0.0.0:*               LISTEN      541308/sems-b2b      off (0.00/0/0)

# ngcp-parallel-ssh all "netstat -ploan | grep heplify"
[1] 15:29:32 [SUCCESS] sp1:22
tcp6       0      0 :::8090                 :::*                    LISTEN      3409461/heplify      off (0.00/0/0)
udp        0      0 1xx.xxx.xxx.11:55003     1xx.xxx.xxx.8:9060       ESTABLISHED 3409461/heplify      off (0.00/0/0)
unix  3      [ ]         STREAM     CONNECTED     527363347 3409461/heplify

[2] 15:29:32 [SUCCESS] sp2:22
udp        0      0 1xx.xxx.xxx.12:50679     1xx.xxx.xxx.8:9060       ESTABLISHED 1350697/heplify      off (0.00/0/0)
unix  3      [ ]         STREAM     CONNECTED     295688769 1350697/heplify

[lmangani]

That's the default Prometheus Metrics exporter port 8090

flag.StringVar(&config.Cfg.PrometheusIPPort, "prometheus", ":8090", "prometheus metrics - ip:port. By default all IPs")

I guess you can replace that with something else, unless you're not scraping it? You can also map it only locally or on a specific interface, if that's helping avoid the collision with SEMS

[ciscospirit]

thanks for your fast reply...

hmm but how? just starting heplify with this argument?

heplify -hs homer.xxx.xx:9060 -m SIP -prometheus :9090

for what is this port needed? is prometheus used by homer or local prometheus? what you mean to map it on a specific interface? for what is this setting?

i also thought about to create the startup script, that it is not allowed to start heplify before sems is running. but not sure if this will solve the problem too.

[lmangani]

This endpoint is used to expose metrics in the prometheus format. They do nothing unless pulled by a scraper. You can use the same setting to modify the :port for all interfaces or a specific one using interface:port Last and perhaps really least, You can also make any other service a dependency for the heplify service if you want it to queue behind it.

[lmangani]

Closing as answered. Feel free to post updates if needed.