Kamailio does not update CSeq from 200OK on

[vinzens]

We've installed latest kamailio from master branch and we see an issue with CSeq numbers.

Scenario is as follow:
A <-> Kamailio <-> B
We do use uacreg to register to B from kamailio. If there is an call from A to B via Kamailio, B challenge DA and from that point the CSeq is not equal on both sides. We do use dlg_manage to track the Difference.

While establishing the call it works until the ACK as reply to 200 OK. The ACK is the first message the CSeq is not updated anymore. In X_log I can see the $dlg_var(cseq_diff) is still there and gives the correct difference. It seems it is not used anymore.

All following requests / responses do have this issue e.G. Bye.

I can provide a wireshark trace, Logs and I can reproduce it in our test lab. Let me know if you need anything.

We do use the latest master version because we had this issue:
#679

unfortunately we were not able to find any commit in 4.4 branch, so we tried master. (would be fine for us, if we would get it run)

cheers
Jöran

[miconda]

The ACK needs to have the same CSeq as INVITE. It is not updated after challenging the INVITE?

Can you attach a pcap for such call?

[vinzens]

The CSeq is updated within the Invite Request. The call gets established and the ACK on the 200OK has the wrong CSeq. After that all messages have wrong CSeq e.G. Bye.

I send the trace to your gmail account. Hope this helps.

[miconda]

Can you try with master branch now? I just pushed a patch.

[vinzens]

I tried the newest version.
In SIP traffic i cannot find any difference, in the Logs where the $dlg_var(cseq_diff) gets printed, i can see the diff gets reset after the ACK on the first 200 OK. The SIP however does not get the incremented CSeq so there are still retransmits.

trace and log are sent via Email.

[miconda]

OK, I will dig in further.

[miconda]

A new patch a pushed, can you give it another try?

[vinzens]

Hi,

i did, but still looks the ACK has wrong CSeq.

sorry

trace atatched

On Tue, Aug 30, 2016 at 5:32 PM, Daniel-Constantin Mierla <
notifications@github.com> wrote:

A new patch a pushed, can you give it another try?

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#752 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AH-W_995O1XvQjk-RtP11esNus3mcB_gks5qlE0ggaJpZM4JoX5S
.

Jöran Vinzens - vinzens@sipgate.de
Telefon: +49 211-63 55 56-21
Telefax: +49 211-63 55 55-22

sipgate GmbH - Gladbacher Str. 74 - 40219 Düsseldorf
HRB Düsseldorf 39841 - Geschäftsführer: Thilo Salmon, Tim Mois
Steuernummer: 106/5724/7147, Umsatzsteuer-ID: DE219349391

www.sipgate.de - www.sipgate.co.uk

[miconda]

Can you give the logs with debug=3 for such case? Also the output of kamailio -v. Yesterday I tested with pjsip and looked ok...

[vinzens]

Hi,

sure. Here the output and the log via Mail.

kamailio -v

version: kamailio 5.0.0-dev6 (x86_64/linux)
flags: STATS: Off, USE_TCP, USE_TLS, USE_SCTP, TLS_HOOKS, USE_RAW_SOCKS, DISABLE_NAGLE, USE_MCAST, DNS_IP_HACK, SHM_MEM, SHM_MMAP, PKG_MALLOC, Q_MALLOC, F_MALLOC, TLSF_MALLOC, DBG_SR_MEMORY, USE_FUTEX, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE, USE_DNS_FAILOVER, USE_NAPTR, USE_DST_BLACKLIST, HAVE_RESOLV_RES
ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144, MAX_LISTEN 16, MAX_URI_SIZE 1024, BUF_SIZE 65535, DEFAULT PKG_SIZE 8MB
poll method support: poll, epoll_lt, epoll_et, sigio_rt, select.
id: unknown
compiled with gcc 4.9.2

[miconda]

Did you compile from git sources? Because the git commit hash is not listed in the version output.

[vinzens]

yes I did.

we do use our own Build System where the Git commit ID gets lost.

the changes which are included in the Build I triggert is that:

#120 (Aug 30, 2016 5:35:09 PM)

tm: uac: Add failover support for local requests — alex / detail
tm: uac: Split t_uac_prepare — alex / detail
tm: uac: Add support for onsend route to local request — alex / detail
tm: Set tcx.branch_index to picked branch before running failure route. —
alex / detail
tm: copy on_reply and on_failure routes to new branch on DNS based — alex /
detail
tm: Copy on_branch_failure to new branch on DNS failover — alex / detail
dispatcher: free memory on shutdown — alex / detail
pv: xavp: log error message when memory allocation fails — alex / detail
core: Enable SIGUSR1 for child processes — alex / detail
mqueu: Convert items to single linked list — alex / detail
tmx: Try to get branch_index for all replies — alex / detail
jsonrpc_s: Return an error when JSON parsing fails — alex / detail
core: Don't bind to IPv6 addresses if auto_bind_ipv6 is disabled — alex /
detail
sl: Reset timeout before sending the message — alex / detail
sl: Convert get_ticks() to the somewhat faster get_ticks_raw() — alex /
detail
pike: fixed rpc array - added array structure when returning multiple —
kelchy / detail
pkg/kamailio/deb: version set to 5.0.0~dev6 — linuxmaniac / detail
modules/sipcapture fixed priorisation X-RTP headers — alexandr.dubovikov /
detail
http_async_client: init PV API sooner — camille.oudot / detail
topos: record routing storage for handling sip responses — miconda / detail
dialog: proper safety test if cseq value in via is larger than hdr —
miconda / detail
dialog: proper refresh of cseq for requests withing dialog after an —
miconda / detail

The One before i Build was with the changes till the 26th.

On Wed, Aug 31, 2016 at 2:38 PM, Daniel-Constantin Mierla <
notifications@github.com> wrote:

Did you compile from git sources? Because the git commit hash is not
listed in the version output.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#752 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AH-W_wDGMz0wB2UwwHtXclCJkr05xhdiks5qlXXQgaJpZM4JoX5S
.

Jöran Vinzens - vinzens@sipgate.de
Telefon: +49 211-63 55 56-21
Telefax: +49 211-63 55 55-22

sipgate GmbH - Gladbacher Str. 74 - 40219 Düsseldorf
HRB Düsseldorf 39841 - Geschäftsführer: Thilo Salmon, Tim Mois
Steuernummer: 106/5724/7147, Umsatzsteuer-ID: DE219349391

www.sipgate.de - www.sipgate.co.uk

[miconda]

Try again with the latest patch above -- in some cases, the required headers might not have been parsed.

[vinzens]

Tested last Version. It works. Great work! many thanks

[miconda]

Welcome, closing this issue.