errors with 5g SA deployment using sa-deploy.yaml

[ukcse19]

I have been looking at your docker solution. when i start core services, as per your instructions, i get errors with configuration in SMF and UPF. see below, any suggestions?
Write out database with 1 new entries
Data Base Updated
Open5GS daemon v2.7.2

01/05 17:49:28.648: [app] INFO: Configuration: '/open5gs/install/etc/open5gs/smf.yaml' (../lib/app/ogs-init.c:133)
01/05 17:49:28.648: [app] INFO: File Logging: '/open5gs/install/var/log/open5gs/smf.log' (../lib/app/ogs-init.c:136)
01/05 17:49:28.714: [sock] ERROR: looks_like_ip(, (null)) failed (../lib/core/ogs-sockaddr.c:602)
01/05 17:49:28.714: [pfcp] FATAL: ogs_pfcp_subnet_add: Assertion `rv == OGS_OK' failed. (../lib/pfcp/context.c:2214)
01/05 17:49:28.714: [core] FATAL: backtrace() returned 8 addresses (../lib/core/ogs-abort.c:37)
/open5gs/install/lib/x86_64-linux-gnu/libogspfcp.so.2(ogs_pfcp_subnet_add+0x84a) [0x7a7b82b94881]
/open5gs/install/lib/x86_64-linux-gnu/libogspfcp.so.2(ogs_pfcp_context_parse_config+0x25cb) [0x7a7b82b8ca48]
./open5gs-smfd(+0xffc3) [0x557f0ebe1fc3]
./open5gs-smfd(+0xf45e) [0x557f0ebe145e]
./open5gs-smfd(+0xfd46) [0x557f0ebe1d46]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3) [0x7a7b82863083]
./open5gs-smfd(+0xf38e) [0x557f0ebe138e]
/open5gs_init.sh: line 102: 60 Aborted (core dumped) ./open5gs-smfd

[herlesupreeth]

Can you post here what modifications did you do after cloning the repository?

01/05 17:49:28.714: [sock] ERROR: looks_like_ip(, (null)) failed (../lib/core/ogs-sockaddr.c:602)

This clearly shows that something was changed in configuration file which is incorrect

[ukcse19]

hi yes sorry. my mistake. for some reason the ue ipv4 addresses were wiped off! its been working since. does your solution work with dockers ok? i mean all sip services such as kamailio rtp get traffic and you have tested it with 2 phones?

[ukcse19]

What sims did you use? I think sysmocom gas to be programmed and IMS files changed prior we use VoLTE!

[herlesupreeth]

I typically use the below command with pySim to program the Sysmocom SIM cards

./pySim-prog.py -p 0 -x 001 -y 01 -s 8988211900000000025 -i 001010123456790 -k 8baf473f2f8fd09487cccbd7097c6862 --op 11111111111111111111111111111111 -o 8E27B6AF0E692E750F32667A3B14605D -a 11111111 -n exp.test --msisdn 9076543210 --epdgid epdg.epc.mnc001.mcc001.pub.3gppnetwork.org --pcscf pcscf.ims.mnc001.mcc001.3gppnetwork.org --ims-hdomain ims.mnc001.mcc001.3gppnetwork.org --impi [email protected] --impu sip:[email protected]

ofcourse, you need to adapt this command with appropriate ADM keys

[ukcse19]

i will try that shortly. i have got sysmoISIM-SJA5-s17 with I believe default MCC 999 and MNC 70. They also sent some k, IMSI, OP values via emails. Can we keep the MCC, MNC, IMSI, k, OP values as they send or do they need to be changed, esp MCC and MNC to 001 and 01? Thanks

[herlesupreeth]

You can use the default PLMN values. But 999 70 is a private network PLMN and using that may result in some restrictions on the phone to connect to certain NR bands. For example, while using 999 70 iPhones in USA are allowed to attach to limited NR bands which are reserved for private networks.

On the other hand 001 01 is a test network PLMN, where the device is allowed to attach to NR bands reserved for public networks

[ukcse19]

thank you Herlesupreeth, I will try shortly. just installing ubuntu desktop on my VM to program sysmocoms!
on the server, I already have fully followed your instructions last night and achieved following. Two phones (samsung A35 and Alcatel 5002u) connected. I get messaged in mme that 2 UEs are connected. I beleive in smf/upf i saw ips being assigned to these two. I did not establish if UEs were successfully registering though! Do we need any changes in devices for registration to be successful. I think it is a pre-requisite before call can take place! When I used kamailio directly with linphone in the past, registered users will appear in 'Location' table via usrloc module i believe in Kamailio.. in this setup, which database/table do the registered devices live? Thanks

[ukcse19]

in all of these domains, did you mean mnc01.mcc001 not mnc001.mcc001. did you have typo here mnc typing wrongly mnc001?? pdg.epc.mnc001.mcc001.pub.3gppnetwork.org --pcscf pcscf.ims.mnc001.mcc001.3gppnetwork.org --ims-hdomain ims.mnc001.mcc001.3gppnetwork.org --impi [email protected] --impu sip:[email protected]

[herlesupreeth]

did you have typo here mnc typing wrongly mnc001??

no, if the MNC is of two digits a zero gets appended at the start for the domain. Therefore, for MNC 01, the IMS domain is ims.mnc001.mcc001.3gppnetwork.org

[ukcse19]

can we achieve video call one way or another. if not using dialer app, can we use linphone or zoiper. if so, what can be the credentials of login.. sip needs username, password and domain. it also helps to give proxy address which is usually kamailio address. thanks

[herlesupreeth]

First, I dont know why S-CSCF is not forwarding. In order to know that I need a fresh pcap for the following scenario:

1. Bring down all the services
2. Start wireshark/tcpdump (no capture filters)
3. Bring up all services
4. Attach two UEs and then attempt a call
5. Stop packet capture and upload here

[ukcse19]

please find attached the pcap file as requested. I also took the docker logs for containers mme, pcscf, icscf, scscf, mme, pyhss and pasted in the word document in the attachment.
Couple of observations:
• MME logs show two devices attaching. At 19:40:26 one of the UE was removed. This was the point when UE made the call. Upon hanging up, two UEs were reconnected.
• Rtpengine showing fatal errors trying to fine xt_RTPEngine module and ip6_tables. Further errors on not being able to initialize ip6tables and then failing on not being able to create KWENET TABLE 0. If we are not using ip6 then not sure why it is pointing to it within rtpengine?
• Pcscf showing number of errors at the start ‘ ERROR: ims_ipsec_pcscf [ipsec.c:652]: clean_sa(): Error sending delete SAs command via netlink socket: No data available’ and warnings with regards to IPSec polices and NATMANAGE. It then produces event_route error.
• ICSCF shows error connecting to 172.22.0.18 port 3875 ‘connection refused’
• SCSCF show same error trying t connect to 172.22.0.18 port 3875
• I note from PyHSS log it only came up a second before both ICSCF and SCSCF were trying to connect so not sure if it was trying too early?

So I ain't too sure yet what these logs mean but I am hopeful you, the expert, will now be able to find what may be missing on my end! there may be some silly issues!! Thank you
docker_open5gs120105.zip

[herlesupreeth]

I see there is only one PDN connectivity request for "ims" APN. I am assuming it because you have created only "ims" APN and have it selected on the phone. "ims" APN should be selected in the APN menu rather you should have "internet" APN as selected

[ukcse19]

both phones are on ims during call. Are you suggesting one phone is not? from the log file, is there anything you are finding totally out of order? What i am trying to establish if we have done everything you expect for this docker solution to work. Does your solution involve UA devices to be registered in some database such as mysql which is the case when we run kamailio and rtpengine! Before call, usually registration needs to be completed from my experience. What do you think I should do as I have to show demo to my boss on Friday!

[ukcse19]

truly happiest day of my life. By selecting internet as APN on both phones and making phone call, it all went through. You are a genious and your work is truly an excellent precedent of your skills within software engineering :). I am trying now 5g and heard that there are lots of services that need to be disabled first. are you able to advise. thanks a million so far