Vodia PBX on AWS - pbxctrl possible memory leak and CPU utilization

[GregV]

Vodia Support,

I have the Vodia PBX running on AWS t2.mirco (1gb memory) on Ubuntu.
Four to five times a week the AWS instance hits 100% CPU utilization and a console message reports Out of memory;

 

The OS kills the pbxctrl process and it restarts the process and everything returns to normal after about 15 minutes.

When this occurs, all active calls are killed and all extensions lose registration. Most handsets take up to 1 hour to re-register unless the users restart their IP Phones.

You can see in the above screenshot the memory is full as well. Memory utilization generally is 300 to 340mb.

Software version: 67.0.6 (Debian64)
Build date: Jul 21 2021 18:01:40

The PBX has 5 active domains with approx 45 registered extensions.

The PBX has only Vodia PBX installed.

During the day the PBX runs at under 10% CPU utilization as shown below.

The spike you see in the screenshot below is when the CPU utilization hits 100% and the OS kicks in the terminates the Vodia PBX process.

 

 

Can you please assist me in solving this problem?

 

Thank you.

 

Greg

 

[GregV]

Here is another process restart I just had this morning in addition to the one above.

 

[DanielA]

We encountered this issue a few weeks back - what version are you running?
Upgraded to 67.0.6 and haven't encountered the issue again.

[GregV]

7 minutes ago, DanielA said:

We encountered this issue a few weeks back - what version are you running?
Upgraded to 67.0.6 and haven't encountered the issue again.

Hi Daniel,

Running 67.0.6
Was running 67.0.5 and has same problem.

Still occurring on 67.0.6.

Greg

[DanielA]

1gig seems a bit on the low side - we we're running with 4GB and it currently sits around 1.4gb
First time it happened to us, the Azure VM just restarted itself and everything came back online.
Second time we had to manually intervene. 
Since upgrading we haven't had an issue (still keeping an eye on it though)

I'd suggest opening a support ticket - support.vodia.com

[GregV]

1 minute ago, DanielA said:

1gig seems a bit on the low side - we we're running with 4GB and it currently sits around 1.4gb
First time it happened to us, the Azure VM just restarted itself and everything came back online.
Second time we had to manually intervene. 
Since upgrading we haven't had an issue (still keeping an eye on it though)

I'd suggest opening a support ticket - support.vodia.com

the memory requirements are relative like anything... given under normal circumstances the instance runs at only 330mb max memory and average 10% CPU, at this point the resources are adequate. In anycase, every-time this occurs its the pbxctrl that is the process at fault and the OS kicks in and kills it. then it restarts and all is good once again.

Thanks Daniel, I'll open a ticket with Vodia, I haven't done that yet.

 

[Vodia PBX]

We have looked at this in mode details. 1 GB of RAM is a little bit too small, the number of extensions in not the main factor for the memory size. Other factors like the number of CDR are also important. What is happening when the PBX gets close to the limit is that other processes that typically run on a AWS instance that are talking care about house keeping are not able to run and from then on things are going South to the degree that the AWS instance is not controllable for a few minutes. We have added also some more logs that provide more detail reports if the PBX gets hammered with requests (on log level 3) so that we can see if we can just cut them off. For now the recommendation is to upgrade to more than 1 GB.

[GregV]

I've upgraded to 4gb memory and have not seen the problem return.

Thank you both for your assistance.

Greg