provided by Star2Billing S.L.

Hi Guys


We have a 4GB, Dual CPU (quad core) Dell 1950 server with dual 15k RPM Raid-1 on Sangoma E1.

We've installed CentOS 5.1 with asterisk and A2billing.

the problem is the A2billing IVR takes at least 4 seconds and I'm not sure what could be wrong. the CPU usage doesn't even reach 1 percent so I'm guessing we have some wrong software or module or need to change something in bios perhaps?

Thanks
Good luck

Before you can address this delay you need to ascertain exactly what is causing it.
If you're having no luck deducing this information yourself then perhaps you should consider paying someone with a lot of Linux + Asterisk experience to work on the problem for you.

funny thing is I've paid a person to install this who claims a good knowledge but he keeps saying your system have hardware problem which I don't believe is the case.

I think must be something to do with asterisk's build for i686 or something that is not well adjusted in compiling.

although I tried to install a2biilling on top Trixbox myself on the same server and it came up with same result.

so I'm kinda blank where to search for the source of this problem. besides I would like to learn what is the issue so I can track similar error later as well.

Thanks for the reply

Without seeing the evidence provided by a hardware diagnostic tool, I too find this to be a dubious conclusion. Which aspect of the hardware does he suspect and what tools/configurations has he used to test his theory? I have experienced problems with Dell hardware previously, but these were very infrequent kernel panics under heavy I/O on a PERC5 RAID controller. I use Linux's md RAID these days on 'standard' controllers on PowerEdge 1950 and 2950 (I think, although I've never seen them) with no problems. I don't think I've ever run a kernel older than 2.6.15 on these servers though; they'll probably work, but I can't vouch for them.
For what it's worth I'm using 64-bit Gentoo. As you have 4GB of memory you really should be using a 64-bit OS too. This applies to both Windows and Linux.
If you were seeing bursts of 100% CPU usage during the delays then that might be a valid thing to check. However I thought you said your system looks idle?
This is not a Linux support forum, so we are veering very much off-topic. With this is mind I shall just give you a brief list of things I would check or tools I might use: kernel boot flags (especially APIC and ACPI)
tail -F /var/log/messages /var/log/mysql.log /tmp/a2billing_agi.log etc etc (try to see what it's doing when the delay occurs)
wireshark or tcpdump (the delays may not be introduced by your machine, eg DNS resolution)
bonnie++ (or some other IO throughput measurement tool)
htop
iotop
latencytop
strace -fF asterisk

very difficult to recompile latencytop on my pc,
the rest of the testing looked normal.
once i could manage to install latencytop i will let u know.

thanks for the reply

I couldn't install latencytop yet since it needs recompile the kernel and I'm not good at finding the flag for latencytop.

however, I could test after stopping syslog-ng and it was extremely faster.

what could be wrong with syslog?

Thanks

Ask the CentOS folks.

very fast reply I not think Centos is for folks. And is teh system more people use this not means is for folks possible teh more easy . The best Solaris or Debian .

anywhere I think more folks is the people of Debian with computers I have my own complains.
BUt is clear DELL is never IBM or HP

Is only a folk with marketing sell COMPUTERS AND SERVER BASED IN WORK IN WINDOWS