|
to naf
Re: Asterisk Google Voice SIP testing and technical discussionsaid by naf:stability report: another error-less night of logs. inbound and outbound still work in the morning. feeling good...
anyone else? All good for me as well. |
actions · 2018-Jun-30 9:44 am · (locked) |
|
naf
Member
2018-Jun-30 10:27 am
said by maxm:new log from last night to this morning, still see SSL error. I tested inbound call and fails too after SSL error.
21:02:10 asterisk start 22:49:05 first SSL error 01:12:33 SSL 02:06:27 SSL 03:58:53 SSL 07:26:41 SSL 08:48:31 got up this morning and tried inbound call, asterisk responded incoming with SIP/2.0 401 Unauthorized 08:49:20 i tried again.. 08:50:50 i did core reload in cli 08:51:05 tried again, same 401 error 08:51:49 i restarted asterisk 08:52:04 incoming succuessful (ignoring the SSL errors for a second, which i believe only effect outbound calls after it reconnects, cause the inbound problems are different) [Jun 30 08:48:31] VERBOSE[19254] res_pjsip_logger.c: <--- Received SIP request (1942 bytes) from TLS:64.9.242.108:5061 --->
INVITE sip:s@192.168.1.187:5061;transport=TLS SIP/2.0
Via: SIP/2.0/TLS 64.9.242.108:5061;branch=z9hG4bK-524287-1---d6ccdcc94b7fc554c8e5c9e2747411b8;rport
Via: SIP/2.0/UDP XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;branch=z9hG4bK-524287-1---768289f9d8608d80a0cd31660c03621c;econt=FJZP4G6R4FIFOGQUYRI
Via: SIP/2.0/UDP XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;branch=z9hG4bK1901590842;econt=2DBTL5NSVXDZB75TIRFCQSADBB7T5MX5JZMOFHUMVANAEVHOOACC6XABO
Max-Forwards: 68
Record-Route: <sip:64.9.242.108:5061;lr;transport=tls>
Record-Route: <sip:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;lr;transport=udp;uri-econt=SPBC3OJEB>
Contact: <sip:+1XXXXXXXXXX@XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;transport=udp;uri-econt=4E62KXKCDVSN4WDTTTPLJYI6VK6AQ>
To: <sip:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX===@obihai.sip.google.com>
From: "1XXXXXXXXXX" <sip:+1XXXXXXXXXX@XXX.XXX.XXX.XXX:5060>;tag=2121373302
Call-ID: lioEg8YOdQ3MEm1E_5277376555914828871
CSeq: 401803 INVITE
Allow: ACK, BYE, CANCEL, INVITE, UPDATE
Content-Type: application/sdp
Supported: 100rel
Privacy: none
P-Asserted-Identity: "1XXXXXXXXXX" <sip:+1XXXXXXXXXX@XXX.XXX.XXX.XXX>
P-Called-Party-ID: <sip:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX===@obihai.sip.google.com>
Content-Length: 552
v=0
o=- 501178585 1530362911552 IN IP4 74.125.39.58
s=SIP Call
c=IN IP4 74.125.39.58
t=0 0
a=ice-lite
a=ice-pwd:dpUYLVi3eUUdxZf+TJzLX1i/
a=ice-ufrag:Z75eSnOUUkASpZ0T
a=group:BUNDLE audio
a=fingerprint:sha-256 16:61:CE:09:B3:82:D2:81:DE:77:DB:B6:62:1C:CB:7E:D0:1B:F3:0B:D4:F7:D2:89:F1:74:35:45:2E:C3:FE:6E
a=setup:actpass
m=audio 19305 RTP/AVP 0 101
a=rtpmap:0 PCMU/8000
a=rtpmap:101 telephone-event/8000
a=rtcp-mux
a=candidate:1 1 UDP 1 74.125.39.58 19305 typ host
a=candidate:2 1 UDP 2 2001:4860:4864:2::58 19305 typ host
a=sendrecv
[Jun 30 08:48:31] DEBUG[20545] pjproject: sip_endpoint.c Distributing rdata to modules: Request msg INVITE/cseq=401803 (rdata0x7f17100dd308)
[Jun 30 08:48:31] NOTICE[20545] res_pjsip/pjsip_distributor.c: Request 'INVITE' from '"1XXXXXXXXXX" <sip:+1XXXXXXXXXX@XXX.XXX.XXX.XXX>' failed for '64.9.242.108:5061' (callid: lioEg8YOdQ3MEm1E_5277376555914828871) - No matching endpoint found
[Jun 30 08:48:31] DEBUG[20545] pjproject: endpoint .Response msg 401/INVITE/cseq=401803 (tdta0x7f1710024be8) created
[Jun 30 08:48:31] VERBOSE[20545] res_pjsip_logger.c: <--- Transmitting SIP response (1180 bytes) to TLS:64.9.242.108:5061 --->
SIP/2.0 401 Unauthorized
Via: SIP/2.0/TLS 64.9.242.108:5061;rport=5061;received=64.9.242.108;branch=z9hG4bK-524287-1---d6ccdcc94b7fc554c8e5c9e2747411b8
Via: SIP/2.0/UDP XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;branch=z9hG4bK-524287-1---768289f9d8608d80a0cd31660c03621c;econt=FJZP4G6R4FIFOGQUYRI
Via: SIP/2.0/UDP XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;branch=z9hG4bK1901590842;econt=2DBTL5NSVXDZB75TIRFCQSADBB7T5MX5JZMOFHUMVANAEVHOOACC6XABO
Record-Route: <sip:64.9.242.108:5061;transport=tls;lr>
Record-Route: <sip:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:5060;transport=udp;lr;uri-econt=SPBC3OJEB>
Call-ID: lioEg8YOdQ3MEm1E_5277376555914828871
From: "1XXXXXXXXXX" <sip:+1XXXXXXXXXX@XXX.XXX.XXX.XXX>;tag=2121373302
To: <sip:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX===@obihai.sip.google.com>;tag=z9hG4bK-524287-1---d6ccdcc94b7fc554c8e5c9e2747411b8
CSeq: 401803 INVITE
WWW-Authenticate: Digest realm="asterisk",nonce="1530362911/59b26dafd9edcf8c82542bf290fc118d",opaque="421418363803a500",algorithm=md5,qop="auth"
Server: Asterisk PBX GIT-master-8990562e28
Content-Length: 0
ha, your asterisk responds to GV's INVITE to tell google *it* is unauthorized. bold. but the "No matching endpoint found" is informative. it just means something in the config is borked, like the identify or whatever (someone who actually knows asterisk will be able to help). can you post your pjsip.conf (with the refresh_token removed obviously)? |
actions · 2018-Jun-30 10:27 am · (locked) |
file join:2011-03-29 Riverview, NB |
file
Member
2018-Jun-30 10:34 am
It means no identify section matched the IP address to an endpoint. You can try using the line functionality[1] which adds some data to the Contact which may or may not work with Google, which removes the need for an identify section. It requires the remote side to preserve the information (which they should according to the RFC) but sometimes don't. [1] » blogs.asterisk.org/2016/ ··· -option/ |
actions · 2018-Jun-30 10:34 am · (locked) |
|
naf
Member
2018-Jun-30 10:37 am
(to contradict myself) but strangely it can find the endpoint at 9:52, with presumably the same config... ? |
actions · 2018-Jun-30 10:37 am · (locked) |
file join:2011-03-29 Riverview, NB |
file
Member
2018-Jun-30 10:41 am
The hostname is giving me a different IP address, and it doesn't appear Google is returning multiple results (the identify section will match on any of the IP addresses returned by a DNS lookup if multiple are returned). If a reload were issued and a new IP address was returned, then subsequent requests would not match it. That's the only thing that comes to mind. |
actions · 2018-Jun-30 10:41 am · (locked) |
1 edit
1 recommendation |
to naf
The SSL error kills inbound calls on our RasPi test machine. The workaround for us has been to issue: pjsip send register gvsip. We built a little cron job to issue the command every 5 minutes and it seems to be stabilizing inbound calls. The register command doesn't kill calls in progress, by the way. asterisk -rx "pjproject set log level 5"
asterisk -rx "pjsip send register gvsip"
sleep 15
TEST=`tail -n 40 /var/log/asterisk/full | grep service-route`
if [[ -z "$TEST" ]]; then
echo "ssl error. restarting..."
asterisk -rx "pjsip send register gvsip"
sleep 15
TEST=`tail -n 40 /var/log/asterisk/full | grep service-route`
if [[ -z "$TEST" ]]; then
asterisk -rx "pjsip send register gvsip"
fi
else
echo ok
fi
|
actions · 2018-Jun-30 10:47 am · (locked) |
1 edit
1 recommendation |
naf to file
Member
2018-Jun-30 10:57 am
to file
said by file:The hostname is giving me a different IP address, and it doesn't appear Google is returning multiple results (the identify section will match on any of the IP addresses returned by a DNS lookup if multiple are returned). If a reload were issued and a new IP address was returned, then subsequent requests would not match it. That's the only thing that comes to mind. confirmed that google does relay the line param back to us in the invite, so lets remove the identify section completely and use that instead, for the dns reasons you state. ill update the recommended config... maxm, can you update the config like this: » github.com/naf419/asteri ··· 326d5d0f and try inbound consistency again? |
actions · 2018-Jun-30 10:57 am · (locked) |
maxm join:2002-10-14 Atlanta, GA |
maxm to naf
Member
2018-Jun-30 11:05 am
to naf
I copied your pjsip.conf and only added the following line for my ip phone. I'll test the new pjsip.conf with identify removed.
[6001] type = endpoint context = from-internal disallow = all allow = ulaw aors = 6001 auth = auth6001
[6001] type = aor max_contacts = 1
[auth6001] type=auth auth_type=userpass password=mypassword username=6001 |
actions · 2018-Jun-30 11:05 am · (locked) |
1 recommendation |
dziny to naf
Member
2018-Jun-30 11:49 am
to naf
My outbound calls started to fail after this change. Not always, just inconsistent. |
actions · 2018-Jun-30 11:49 am · (locked) |
|
naf
Member
2018-Jun-30 11:59 am
said by dziny:My outbound calls started to fail after this change. Not always, just inconsistent. probably unrelated? post or pm me log of sip message traffic so we can see if its the double-clutch registers or something new |
actions · 2018-Jun-30 11:59 am · (locked) |
naf |
to AllThumbs
said by AllThumbs:The SSL error kills inbound calls on our RasPi test machine. The workaround for us has been to issue: pjsip send register gvsip. We built a little cron job to issue the command every 5 minutes and it seems to be stabilizing inbound calls. The register command doesn't kill calls in progress, by the way. that log shows something different than what im calling the double-clutch register. in the double-clutch case, the first REGISTER received an OK response, but 5-15seconds later the connection drops (SSL_ERROR_ZERO_RETURN) and asterisk automatically re-REGISTERs which also returns a successful response, but the subsequent outbound calls are rejected with 421. in your log, the REGISTER response is not a success. asterisk already handle the 603 response by retrying at the specified interval, and it could handle the 421 in a similar manner if so configured, although it would be better to resolve whats causing the 421 instead. no scripts necessary in either case. p.s. need to get that verbose message traffic in your logs to see the actual sip messages. i keep having to issue this on every restart with asterisk13: pjsip set logger on |
actions · 2018-Jun-30 12:14 pm · (locked) |
xekon join:2017-09-15 Hoodsport, WA 1 edit |
xekon to naf
Member
2018-Jun-30 12:32 pm
to naf
I have been working on getting this installed/built from scratch.
Edit: resolved now. |
actions · 2018-Jun-30 12:32 pm · (locked) |
|
1 recommendation |
to naf
said by naf:said by dziny:My outbound calls started to fail after this change. Not always, just inconsistent. probably unrelated? post or pm me log of sip message traffic so we can see if its the double-clutch registers or something new Never mind, it was fail2ban acting up and banning of one the IPs google is using (136.22.68.172). The clue was suddenly one way sip message traffic. |
actions · 2018-Jun-30 1:19 pm · (locked) |
|
to naf
said by naf:said by AllThumbs:The SSL error kills inbound calls on our RasPi test machine. The workaround for us has been to issue: pjsip send register gvsip. We built a little cron job to issue the command every 5 minutes and it seems to be stabilizing inbound calls. The register command doesn't kill calls in progress, by the way. that log shows something different than what im calling the double-clutch register. in the double-clutch case, the first REGISTER received an OK response, but 5-15seconds later the connection drops (SSL_ERROR_ZERO_RETURN) and asterisk automatically re-REGISTERs which also returns a successful response, but the subsequent outbound calls are rejected with 421. p.s. need to get that verbose message traffic in your logs to see the actual sip messages. i keep having to issue this on every restart with asterisk13: pjsip set logger on Here you go... |
actions · 2018-Jun-30 1:43 pm · (locked) |
1 edit
1 recommendation |
naf
Member
2018-Jun-30 2:00 pm
said by AllThumbs:said by naf:said by AllThumbs:The SSL error kills inbound calls on our RasPi test machine. The workaround for us has been to issue: pjsip send register gvsip. We built a little cron job to issue the command every 5 minutes and it seems to be stabilizing inbound calls. The register command doesn't kill calls in progress, by the way. that log shows something different than what im calling the double-clutch register. in the double-clutch case, the first REGISTER received an OK response, but 5-15seconds later the connection drops (SSL_ERROR_ZERO_RETURN) and asterisk automatically re-REGISTERs which also returns a successful response, but the subsequent outbound calls are rejected with 421. p.s. need to get that verbose message traffic in your logs to see the actual sip messages. i keep having to issue this on every restart with asterisk13: pjsip set logger on Here you go... Ya, those 603 Rejected Retry-After: 600 responses just mean you've been naughty and tried REGISTERing a bunch in succession and you need to just wait the 10 minutes (asterisk will do this automatically) before retrying. Those 421 Extension Required Require: path responses are because the REGISTER doesn't have the "Supported: path, outbound" like it should, either because the config'd registration doesnt have 'support_path=yes' and 'support_outbound=yes' or because i missed a spot in the code that is needed to add them when using the cli register, or because you're not up-to-date with the current code? |
actions · 2018-Jun-30 2:00 pm · (locked) |
naf 1 edit |
to xekon
said by xekon:Was there a way I could have git cloned that would not be reporting the asterisk version as GIT-master-2fd0a875d4
I assume its ok for me to be asking for help here for setting up the GVsip code, if not then let me know and I can start a thread somewhere else. Thank you. yes, this is the right place: cant report test results if you dont get it setup, right? "GIT-master-2fd0a875d4" is fine: asterisk's get-version-from-git script just hardcodes the branch name as master without looking, so that's just how its going to show up [also, i stole your one-liner clone command and put it in the intro post] |
actions · 2018-Jun-30 2:30 pm · (locked) |
xekon join:2017-09-15 Hoodsport, WA |
xekon
Member
2018-Jun-30 3:10 pm
Thanks! one more question, I am assuming because it says master, thats because gvsip is a fork of master? That would mean your gvsip branch is based off Asterisk 15 correct? If not, what branch of Asterisk is gvsip based on? I know plenty of people are still using Asterisk 13 as well, so I figure I better ask.
I see your opening post says:
download the gvsip branch as a patch and apply it against asterisk 13.21.1
so maybe asterisk 13.21.1 is what gvsip is forked from? |
actions · 2018-Jun-30 3:10 pm · (locked) |
2 recommendations |
naf
Member
2018-Jun-30 3:18 pm
it just says master, cause thats whats hardcoded as MAINLINE_BRANCH in build_tools/make_version
but yes, i forked their master branch. but there's not any conflicting changes between 13 and master, so you can apply it against either. |
actions · 2018-Jun-30 3:18 pm · (locked) |
maxm join:2002-10-14 Atlanta, GA 2 edits |
maxm
Member
2018-Jun-30 4:07 pm
naf:
Here is something interesting:
Inbound is working fine even after SSL error now.
Outbound will fail after SSL error, but after I take a incoming call, it will work again...
I sent you a new log
--edit
Scratch that, I just tested again. Incoming works, but outgoing still fails after SSL error |
actions · 2018-Jun-30 4:07 pm · (locked) |
jebs join:2018-06-30 North Vancouver, BC |
to WhyADuck
sorry not trying to ruffle any feathers, but i'm just trying to figure out how to remove the 48 dial plan deleting just 48 from the .conf file doesn't seem to work
any help would be appreciated
thanks |
actions · 2018-Jun-30 4:41 pm · (locked) |
|
WhyADuck
Premium Member
2018-Jun-30 5:05 pm
jebs, did you remember to reload the configuration after you made the changes? Please start a new thread if you need any more help with this. |
actions · 2018-Jun-30 5:05 pm · (locked) |
maxm join:2002-10-14 Atlanta, GA
1 recommendation |
maxm to naf
Member
2018-Jun-30 5:29 pm
to naf
I tried fooling around with pjsip register and found following:
pjsip send unregister gvsip seems to giving me "SSL 6 [SSL_ERROR_ZERO_RETURN]" reliably.
Then when i try pjsip send register gvsip, It will show rejected first then after 60s it will register. |
actions · 2018-Jun-30 5:29 pm · (locked) |
xekon join:2017-09-15 Hoodsport, WA |
to naf
I have the GVsip git clone asterisk built and installed as well as freepbx, and my extensions are registered. I think all that is left is to implement pjsip_custom and modify rtp.conf and extensions.conf according to the github page: » github.com/naf419/asteri ··· ee/gvsipIs the posted pjsip config on the github page still the recommended setup, or has anything changed? |
actions · 2018-Jun-30 6:35 pm · (locked) |
c4pt00 join:2018-07-01 New York, NY |
to xekon
its the media transport dropping one side of the stream check your firewall for inbound outbound ports for the Real Time Protocol* 10000-20000? check other relevant ports, with a network top like "jnettop" or pfTop
Outbound calls however, The other end can hear me, but I cannot hear them. |
actions · 2018-Jul-1 12:10 am · (locked) |
c4pt00 |
to xekon
this is a port or NAT problem |
actions · 2018-Jul-1 12:11 am · (locked) |
c4pt00 |
to jsolo1
check port, NAT settings, check other settings in your config |
actions · 2018-Jul-1 12:11 am · (locked) |
c4pt00 |
to TheTechGuru
@TheTechGuru see the post about one sided media streams with the media transport |
actions · 2018-Jul-1 12:41 am · (locked) |
|
naf to maxm
Member
2018-Jul-1 7:33 am
to maxm
said by maxm:one observation:
The SSL error time is not random. It's the exact time when the registration time is set to expire.
Asterisk refresh the registration 10s before expiration. Sometime google still let it expire after successful refresh.
Is it possible to refresh registration sooner? If you want to try it, increase the value of REREGISTER_BUFFER_TIME in res/res_pjsip_outbound_registration.c |
actions · 2018-Jul-1 7:33 am · (locked) |
1 edit |
to naf
I've noticed that people using naf Asterisk or Simonics GVSIP still have 'chat' available (switchable) in their GV accounts but people using Obi200's are fully converted over to GV-SIP. Could this be why you're seeing issues with Asterisk? Does everyone maybe have to 'fully' convert? Edit: reduced number of questions. |
actions · 2018-Jul-1 12:07 pm · (locked) |
GIvic join:2018-06-30 Mountain View, CA |
to xekon
That is the current latest for testing. Last commit is, » github.com/naf419/asteri ··· 326d5d0f |
actions · 2018-Jul-1 5:51 pm · (locked) |