-
Posts
1,766 -
Joined
-
Days Won
4
Content Type
Profiles
Forums
Resource Library: Monster UI Apps for KAZOO
Events
Downloads
Posts posted by mc_
-
-
Hi Ash,
First, welcome! Glad you're here!
I would start by looking at the SDP in the A-leg's INVITE to see codecs offered, as well as what codecs are then offered to the B-leg from FreeSWITCH. Could be a transcoding issue or misconfigured codec selection.
Payload type 0 is typically PCMU but you're log shows PCMA being used. Perhaps look into what codecs FreeSWITCH thinks are loaded? `fs_cli -x 'show codecs'` should list `codec,G.711 ulaw,CORE_PCM_MODULE` among others.
-
Sorry, yeah, I was unclear. It's a code change in kz_hooks_util.erl to prevent the crash. Would need to be a PR to the 4.3 branch on the kazoo repo. I'll try to get to this but if you (or anyone!) want to get it first, that'd speed up the process.
-
Quick guess is that gproc throws an exception when a binding already exists. A patch to use gproc:ensure_reg/1 would probably address this (in kz_hooks_util)
-
You can disable call_waiting on the device or user config: https://docs.2600hz.com/dev/applications/crossbar/doc/users/#call_waiting
-
@Shah are you creating one connector doc per PBX or one doc for all the PBXes?
-
Correct, "pvt_attachments" is a KAZOO-added thing for the metadata. "_attachments" is the Couch key for binary data attached to the document. No "_attachments", no recording data
-
FreeSWITCH caches the recording so it could be your cache didn't time out yet, but please do check. We don't want to be redundant in saving the binary!
-
Well, you would need to check the document in couch to be sure. It should contain all the metadata about the recording but should not have an '_attachments' key with the binary data.
-
Multiple instances of each app are definitely encouraged and supported!
Unfortunately I cannot think of why multiple instances might affect saving media. Would require more investigation for sure.
The basic idea is, when KAZOO is ready to store the recording, it builds a URL to a media_mgr instance for FreeSWITCH to stream the recording off the server to the media_mgr server. Once received, the process in media_mgr will 'save' the recording which should use your storage plan's config to save to the configured backend.
Hopefully you can use the mp3 filename to key in on logs on nodes running ecallmgr and media_mgr.
-
So the proxy URL built is to stream S3 attachment to kazoo media_mgr node, then FreeSWITCH is given this proxy URL to fetch the media from (instead of FreeSWITCH going direct to S3).
As for the timezone issue, I'm not sure what's causing that crash when getting unix epoch for your timezone. I can manually run the code fine with 'Asia/Jerusalem' as the timezone.
Can you find that message's metadata in the MODB and see what the 'timestamp' key is set to?
-
How are you configuring call recording to occur? By configs on account/user/device or explicitly in the callflow ?
-
Yup so grep that kz_att_http PID '0.10111.234' to see more info.
But it looks like you're using the HTTP handler to put your recordings in AWS; perhaps you've misconfigured the storage plan to use http instead of s3?
-
In this case, the app `media_mgr` is setting up that URL to feed to FreeSWITCH for getting the recording off FreeSWITCH. Once received from FreeSWITCH, media_mgr will attempt to store the media to the configured backend.
So check any nodes where media_mgr is running (ensuring log level is debug) and you should see kz_att_* logs related to the attempt to upload the file to the storage backend.
-
You'll need to check the KAZOO logs for that mp3 filename to see what kz_att_* logs might have to say.
-
This has been part of 4.3 since 4.3.90: https://github.com/2600hz/kazoo/releases/tag/4.3.90
So make sure you have a version >= .90.
But, more importantly, you continue to show examples using "sequential". That means "in the order they appear in the gateways array".
-
You want to set `gateway_strategy` to `random` on the top-level of the document (not per-gateway). See the schema at https://docs.2600hz.com/dev/applications/crossbar/doc/resources/
Sequential will dial the gateways in the order they appear in the array; random will shuffle the array and dial in that randomized order. Assuming the shuffle is "random" enough, you should see a roughly even balance of which gateway is dialed first.
-
@Pargat sounds like, unfortunately, a lot of spam and malicious traffic originates from various countries; operations has blocked large swaths of IP blocks to combat it and you appear to have been caught up in that.
We're working on getting a mirrored host up so that folks like you can access the repos without needing to resort to a proxy or VPN. More to come!
-
Not aware of any restrictions (asking around). Can you try fetching that URL with wget or cURL and see what they say?
-
You cannot, SUP is only for the cluster administrator. What are you trying to do as a sub-account that you need them to use SUP for?
-
Look at the flag for 'gateway_strategy'; can be 'sequential' or 'random'. Then you can define one resource with 3 gateways and shuffle between them.
With resources of the same weight, there is no expected or implied order; its kinda up to the system and how things are fetched from the database.
-
Can you capture the logs from Kamailio and KAZOO on a REGISTER attempt that should have failed but succeeded?
-
@fmateo05 hope its more of a temporary break But if not, wish you well and thank you for all you've contributed! You're always welcome here.
-
Appreciate it @esoare !
-
Hi @naveed6865 can you please start a new post (it appears like you're thread-jacking unintentionally). Please use the search facility as well as I believe this question has answers. If not, please make a new post! Thanks
Codec / payload one way audio issue
in Product Discussion
Posted
Always nice to get a win!