webclient 16.0.4.493 - first seconds for a call between 2 webclients are missing

[dsi-infologic]

Hello
first , the new chrome extension is amazing ! congrat !

just upgraded our test 3CX to the lastest realease , we discover during our tests an issue during the first seconds of a call.
call between 2 webclients ( same 3CX , same VLAN ) : when I take the call from a colleague, the first seconds ( between 2 and 4 ) are blank ( we loose the first and second "Allo") , then all ok
call between a windows app and a webclient : voice ok from the beginning of the call
call from outside to a webclient : voice ok from the beginning
call from a physical phone tp webclient : ok

so , all seems ok expect the commnication between 2 webclients on the same 3CX.
With the chrome plugin, the issue seems to be a little most bigger ( 1 or 2 second more )
note : the keyboard button is "grey" during the issue and comes available as soon as the voice comes.

in our case, PBX delivers audio and codex are PCMA then G722.

I do not found error on our network ( same vlan , same switch , DSCP ok ) , so I expect a slowness during the communication negotiate at the beginning.

3CX version 16.0.4.493
Chrome version : 79.0.3945.88 (Build officiel) (64 bits)
tested with an without incognito mode

BR
Stéphane

 

[RemoMeyer]

We have the same issue with SNOM Phones (M700 DECT + M65/M70) and with Webclient.

At least 4 or 5 Clients of us have this problem, too. We can't until now figure out waht is that

 

[dsi-infologic]

my SNOM is D735 , but seems not impacted.

 

[JohnS_3CX]

Provide some more details as per this post:

https://www.3cx.com/community/threads/information-to-provide-when-requesting-help.67558/

Include some additional info as to how your network is setup, this sounds a little bit like a delay on the network when the client establishes a connection with the PBX

 

[dsi-infologic]

Hello John,

• CX Version, enterprise 16.0.493
• Server OS, Debian 9
• Is the 3CX Server Hosted and where : 3CX is one VM on the VoIP VLAN in our datacenter.
• IP Phone Make/Model/Firmware : chrome extension or 3CX windows app
• Provisioning Method: Local / VPN / STUN / SBC : Local
• Trunk Provider or Gateway Make/Model : OVH/SEWAN but not related in this case
• Has the Firewall Checker passed: YES
• Are custom Phone Templates being used: YES ( for contacts lists )

3CX productive server is on VOIP vlan and users are on wifi / Lan vlan ( all on the same site with a latency < 1ms ). We have a firewall in order to do the gateway between each vlan. logs checks and nothing is dropped.

for the 3CX test server ( standard 16.0.493), users and server are on the same vlan , same issue.

 

[JohnS_3CX]

Thanks,

Also please share the following:

• Server CPU Model
• Number of CPU cores assigned to VM
• Amount of RAM assigned to VM
• What virtualization environment you use (VMware/Hyper-V/etc..)
• How many users are logged in simultaneously (or on average)

 

[dsi-infologic]

• Server CPU Model : Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz
• Number of CPU cores assigned to VM : 4
• Amount of RAM assigned to VM : 8Gb
• What virtualization environment you use (VMware/Hyper-V/etc..) Vmware 6.7
• How many users are logged in simultaneously (or on average) : 220

 

[JohnS_3CX]

You may be a bit under-spec'd for the size of your installation. The recommended CPU for this size of machine will have almost double the performance per core, plus 2 more cores for multithreading.
https://www.3cx.com/docs/recommended-hardware-specifications-for-3cx/

So I would recommend as a test scenario, to increase the vCPU cores to 6, and shut down any other VMs the server may be running to free up resources and see if you still get the issue.

I believe VMware also allows you to reserve the resources the VM uses, so that no other VMs will compete for CPU cycles and or memory and "steal" processing power.

 

[dsi-infologic]

Hi John,

We have two 3CX servers with the same vm cpu/ram/disk

• one productive , enterprise license and +200 users
• one test , standard with less than 10 test users

1°) we also have this issue on the test server
2°) we don't have the issue at all with the windows app (so luckily no impact on our current productive users ) on both test and productive servers.

So I think the "charge" is not the right way, and most a matter of "chrome extension".
I will increase the cpu as requested anyway in order to be on the recommanded specifications.

Thanks

 

[JohnS_3CX]

Ok lets test with 6 vCPUs and see if it helps.

Do you have recordings enabled by the way? (ie. when one webclient extension calls the other)
If so, you can disable recording on 2 of the extensions then make a few calls between them to see again if there is any difference.