Your wrong with that statement, but now I understand why you dont help much with hosting provider…
2 Likes
I understand you are frustrated with the issue you are experiencing, but I think you are wrong here and in the end of the game you are loosing out here…
If you are hanging out here frequently, you will see that around every 3 months comes around a similar CPU/Memory/Audio post, sometimes the OP claims it’s a PJSIP fault, and sometimes a VPS issue.
Every. Single. Time. the developers, asked for more information, and somehow it either turned into OP blaming Sangoma, or the user made changes in the meantime and refused to reproduce it again to be able to provide the requested information. (It works now. Why should I change back to PJSIP?)
Developers are not magicians (well sometimes they are?) they can only fix or improve something when they know what it is. You can’t demand and expect from them to “FIX IT!” without actually cooperating with them…
As you are aware, Andrew is the lead developer at Sangoma/FreePBX, and as he stated:
Sangoma employees do not have to participate in these forums at all.
but he chose to do so, he’s here on daily basis to help and improve the project. If you are hanging out here frequently you would witness how much time and advice he shares to help users like you on the community here.
He could’ve told you. “Yeah, that’s an issue, please open a paid support ticket and we’ll figure it out” But instead he asked you guys for information so he can pass it on to his team to take a look at, to potentially apply a fix.
Finally, there have been many times where a poster have asked for help with a script, of something they are trying to accomplish - in some cases, what they were trying to accomplish is already part of (or partly part of) a commercial module. Many, many times have the Sangoma developers assisted with getting said script to work.
No, they are not going to share with you the source code of commercial modules, but they are all here to help, you and the rest of the community.
3 Likes
My point is: When 1+1 = 2, why trying to say 1+1= Probably not 2 ?
I think I pointed a problem clearly with all the proof… the CPU Peak come from the cronjob, maybe I didn’t point the right files? Maybe that file is supposed to do nothing on the system, BUT when the cronjob is enabled and every minute the problem occurs… When the Cronjob is disabled we have no problem…
I will be happy to show them live with access to the system if they want to do the test himself, I don’t care, but an answer like We dont share our secret is not the right answer here…
How many CPUs (cores) does this thing have?
Host has 64 Core (2cpu 32core)
No, not the host. We’re not looking at the host, we’re looking at the VM. How many cores/cpus did you give the VM?
For the test its 1vcpu,
And the other 20 VMs? Are they all the same? Is this test VM the exact same as the other 20 VM’s or the ones experiencing the dropouts? So far everything you have posted is showing they all have 1 core to them. That is the problem.
I would never put a system for more than 3-5 users on a 1 core VM with current versions of FreePBX. There is too much stuff running now to support current needs and technology for everything that the system is doing. That is before you even start making calls and adding call volume, recordings or anything else that will increase the load.
So far we have one comment that this is a “bug for hosting providers”. How is that even possible? Seriously? Next we have a snide comment about how there is no support for hosting providers because it would be against Sangoma’s interest. That is also BS.
The problem here is poorly configured VMs that aren’t setup for the basic load FreePBX needs to run. This is no different then a hosting provider putting up a cpanel server with 1 CPU and loading 100’s of hosts on it to run as a shared host system. Yeah, it’s going to have issues.
Just for fun too, you’re both Sangoma partners right? You’re not actively marketing yourself as a FreePBX host/resale without having the proper agreements in place right?
No you don’t have proof. You have a changed result. No proof was ever shown in any way that any developer could ever troubleshoot.
Do what was asked. That never happened. No actual results showing any details of the PHP process using up the CPU was documented here.
What’s that???
The test VM has no Call on it And I can see the CPU Spike… I can put 8 cpu on it if you want and we will see the same cpu spike (I tested)
Sure. Do that. 2, 4, 8 doesn’t matter.
Way too small to read when I was on my phone.
So, now you have a single instance of it showing the spike. Great.
Now open htop back up and hit F4 and type php, to filter the view. Then do a good screen recording so Sangoma can see what is happening.
At that point, you may want to open an actual support case up and give them SSH access to the unit so that they can see it and work on it.
1 Like
One work around I have used in a similar situation is to add a random 1-60 second delay before the execution of the script on each machine, it doesn’t fix the underlying problem, it just prevents the on minute peaking
sleep $[ ( $RANDOM % 59) + 1]s; yourcommandline
3 Likes
I do this also with automation on systems for things like certbot or dnf-automatic. I don’t need every system reaching out at exactly the same time to renew certs or check for updates.
Yes that’s what I will do on system that need some of the module that look to cause the problem.
Otherwise the fix for us now its to disabled practically all the cron job…
Hello Andrew,
Again, I am positive and not attacking anyone. Please understand.
What I Expect.
Well, I expect a honest conversation on a problem. If you don’t want to tell the fix, than you can say that. Off course I will understand.
Just wanted to tell that there is a hosting problem with more concurrent FreePBXs-Distro on one host, and you are saying that this is fixed. That’s all I have dreamed off.
Thanks
We have tried the sleep command with random in cron’s. Perfect solution. I wrote it in a earlier post.
But when upgrading a Distro FreePBX, it was set back to original settings, or a notification that files where adjusted.
Anyhow,
It should be fixed in the distro, but it that is not easy to achieve this.
A fix for what. There is no valid information yet known except for one single screenshot of one process.
1 Like
OK so far your contribution has been “I have this too”. You have yet to describe your setup in any way. There is nothing so far to show that you are having the “same problem” in the “same way”. So to have an honest conversation on your problem, you need to describe your setup and actual problem outside of “me too”.