Dropped calls with same number of bots (~25) on weaker and more powerful hardware.

I work for a company that provides remote agents and bots. The auto dialing occurs on the client’s servers and we just receive the calls on our server.

We use a vicidial image with asterisk (recordings on) and run agi bots on the same server.

Now when the agi bot count goes up to around ~28 bots we start experiencing distortion and dropped calls despite having tried and tested a shared and dedicated server both from hetzner (details listed below).

I also suspect that this starts happening when the pace of auto dialing is increased from the client end.

I am new to working with asterisk and not very well versed so I’d like some help as to why this is the case and if there is any way to fix this. We have also been considering a bare metal server and if that would increase call quality and decrease dropped calls. But that is a pain to set up and does not have the features that cloud machines do on hetzner so I’m doing my research.

Any help is appreciated.

AX41-NVMe
AMD Ryzen™ 5 3600 Hexa-Core “Matisse” (Zen2)
64 GB DDR4 RAM
2 x 512 GB NVMe SSD (Software-RAID 1)
1 Gbit/s bandwidth
36 Euro

CPX42 - Shared
8 vCPU
16 GB
20 Euro

CCX23 - Dedicated
4 vCPU
16 GB
160 GB
24 Euro

Asterisk 18.26.2-vici

On Thursday 22 January 2026 at 17:46:06, pjmommy wrote:

… we start experiencing distortion and dropped calls despite having tried
and tested a shared and dedicated server both from hetzner (details listed
below).

We have also been considering a bare metal server

CPX42 - Shared
8 vCPU
16 GB

CCX23 - Dedicated
4 vCPU
16 GB
160 GB

Since neither of these are bare metal servers, what do “shared” and
“dedicated” mean?

Any Virtual Server is going to be hosted on some large physical machine
alongside other customers’ Virtual Servers, and is sharing the resources of
that physical machine with the other Virtual Servers hosted on it.

So, what do Hetzner mean by “dedicated” for a Virtual Server?

Antony.


Ich habe gerade ein besonderes Bier getrunken.
Jetzt habe ich kein Bock mehr :frowning:

Shared servers are virtual servers but with other people running their machines on it too. So resource spikes might affect our workload.

Dedicated resources have no such affect.

Also the cpu quality varies.

What you are doing is like ordering an engine replacement in your car because of a clogged fuel filter. You don’t know anything about what’s going on until you start turning on verbose debugging and looking at the logs. So, instead of diagnosing the problem you think it’s easier to just gut and replace everything in the engine compartment. Sigh.

In addition, your Asterisk version is old. Take a look at the following recent post here:

Taskpool: A Hunch Turned Performance Improvement ⋆ Asterisk

“…Early reports have shown that this has improved CPU usage a noticeable amount (anywhere from 20-30% less CPU usage) and solved performance issues for people…”

I’d wager if you setup a current version of Asterisk under Debian 12 at Hetzner you wouldn’t experience this problem. At least, not until you had much more activity.

1 Like

Sounds like you’re using a Virtual Server & a Dedicated together as a “Cluster” This will not work you need them both on Dedicated Servers. For reasons already explained, I’m sort of a ViciDial expert you an reach me at mejorleads @ gmail