vicidial.org

Hello,
Server Specs:
E3v6 Processor
8 CPU @ 3.8 GHz
32GB RAM
2 TB HDD SATA
All servers in cluster are of same configuration. Cluster setup is 1x DB & Web Server and 4x Dialers.

I am used to do scratch installation on centos 6.9 64bits for building clusters and they work fine for years without any trouble. This time i have tried creating cluster with vicibox iso following the wizard. I had to change hostnames for all servers i had to overcome the duplicate id error while installing vicibox for telephony servers. I selected the database installation and Web server installations on database server and did an "N" for the rest i.e. Telephony server, archive server etc. On the remaining servers i did an N for database server and Y for telephony servers. after all servers were done they were rebooted. As soon as the servers were backonline i logged in to the web interface and saw all dialers and database was automatically there but database shows 100% load on web interface while the rest of the dialers look fine. when i do a TOP on shell i dont see 100% CPU or any utilization which could be abnormal. can anyone please help me out. One more thing i also want to setup webRTC on this cluster so is it enough to run the cetbot thing on database only? or do i have to run on every dialer aswell?

1) You stated something in there about an .iso installer but didn't mention a version. You also have reference to "scratch install", but have not provided a link to the instruction set used. This information is often very useful in understanding your situation.

2) You have mentioned "database load 100%" and "database shows 100% load on web interface" which I presume to mean the value shown when you click on "Reports" and are shown a list of servers. This value is not actually a "load" per se, but a calculation of a value based on an assumed calibrated maximum number of processes. However, this value (when uncalibrated) has zero meaning. Average Server Load, which is available in the CLI, shows the number of cores busy (on average) in the last 1/5/10 minutes. This number should never exceed the actual core count. For safety (peace of mind) it should never really exceed 50% of the core count except perhaps for the occasional peak moment.

3) If you want that "Load" value to show a usable number, you'll need to calibrate the data collection service for it. This, of course, is where the installation method would come in, which is missing.

Sorry i forgot to mention the iso version is vicibox 8.1 latest version available on website

samadsaeed wrote:Sorry i forgot to mention the iso version is vicibox 8.1 latest version available on website

The LATEST version of Vicibox is 8.1.2 (not 8.1). Versions are important all the way to the last digit.

Sorry for that again. But now when we have the version correct can u please help me pin point the issue as when i do top on database shell i dont see any abnormal load

samadsaeed wrote:Sorry for that again. But now when we have the version correct can u please help me pin point the issue as when i do top on database shell i dont see any abnormal load

that's not a problem. top shows load correctly. means you don't have an abnormal load. if you want the "reports" page to show Not an Abnormal Load you need to calibrate the data collector for that role's expectations. But that in no way means anything is wrong just because that value is not calibrated correctly.

Thanks william. Can u guide me on how to calibrate the reports to show it correctly as i understand its not an issue but still when i do scratch installation and manually cluster the whole thing it works perfectly. I follow ray solomon's scratch installation manual which i guess isnt available online anymore but i have it on my hard disk. It has asterisk 1.8 but i install 11.22 thats the only change. But the main thing is the reports section shows accurate loads when everything is ready. Fixing this for vicibox would be a great help sir.

One more thing.. for webRTC do i have to run certbot on all servers or only database is enough. I want to point the whole cluster to a single domain.
I have run certbot on database and completed all steps. I created a phone on dialer 1 but its not connecting. I added the websocket URL generated by certbot on all servers including the dialers. when i removed the websocket url the phone still doesn't register but error message changed from disconnected to Reg. Failed. Can u please guide me what i'm missing. I have many standalone vicibox 8.1.2 installations up and running with no issues at all along with webRTC. its just this cluster thing is getting a bit confusing. your help would really be handy at this moment. Thank u again.

William can u please help me in getting this webRTC thing working for this cluster it would be a great help. I have a client waiting to go live sir. I would be extremely grateful to u. i would really appreciate if anyone whose has a solution for this could write back a solution or a hint for me. I would really appreciate that sir.