Xen PVM and FXO/FXS card is it reliable?

After installing FreePBX 13 on a Xen PVM and connecting it to my PSTN via a SIPURA 3000 box, I’ve come to the conclusion that FreePBX works great under Xen and will do everything that I need, but the SIPURA 3000 is just not as reliable as I would like.

The SIPURA did well enough for testing and setup, but now that I’ve rolled this into production (OK, it’s my home, but it still needs to work), I’m finding that the SIPURA will stop transmitting audio or lockup or some other unexpected and very frustrating thing.

I’d like to switch to one of the FXO/FXS cards and I’ve seen documentation on getting them to work under Xen PCI pass through. I’ve also seen some indication that due to timing issues etc. The result might not be all that good.

Does anyone have any recent experience setup? Everything I’ve seen is 5 to 8 years old. I’d like to think things are working much better now.

That’s a complicated one , it depends on your hardware/motherboard/CPU’s, post the output of

cat /proc/cpuinfo

on domain 0

The dom0 has plenty of power. Two quad cores with hyperthreading and 96G of memory. Here’s the /proc/cpu (only the last entry)

processor : 15
vendor_id : GenuineIntel
cpu family : 6
model : 44
model name : Intel® Xeon® CPU E5620 @ 2.40GHz
stepping : 2
microcode : 0x13
cpu MHz : 2394.060
cache size : 12288 KB
physical id : 0
siblings : 8
core id : 10
cpu cores : 4
apicid : 21
initial apicid : 21
fpu : yes
fpu_exception : yes
cpuid level : 11
wp : yes
flags : fpu de tsc msr pae mce cx8 apic sep mca cmov pat clflush acpi mmx fxsr sse sse2 ss ht syscall nx lm constant_tsc arch_perfmon rep_good nopl nonstop_tsc pni pclmulqdq monitor est ssse3 cx16 sse4_1 sse4_2 popcnt aes hypervisor lahf_lm ida arat dtherm
bugs :
bogomips : 4788.12
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management:

You should be fine if vt-x and vt- d are enabled in your bios

So the issues I’ve heard about are mostly about horsepower of the dom0? That good to know.

Now I’ll get a few people ticket off at me by asking about the clone cards and their reliability. Since this is a home system, I’m not inclined to spend what it would take for a brand new card from the digium or Sangoma. I’ve been looking at used cards on ebay, but I need a pci-e card and those are still a bit much used. That makes the clone cards very attractive.

Not really to do with horsepower, it is to do with whether your cpu and your motherboard support the pass-through of your PCI bus, you NEED both VT-X AND VT-D to have half a chance with any virtualization.

As to cheap shit I commented on this thread:-

Be careful of false economies, for example your machine is a true beastie but will probably cost you way north of $1000 a year in power just to keep it on.

Not sure what this beasie is costing me to run, but I’m willing to bet it is less than half of what it cost me to run its predecessor. In fact, I’d be willing to bet just the disks on that one cost me more than this entire system.

However, good point, an unreliable card is probably not worth it. I seem to have stabilised my sipura 3000, so I have time to wait for a good deal on a used card.

1 Like

After rebooting my dom0, I check the bios and the virtualisation stuff is on, however I can’t see the flags I’m expecting in the /proc/cpuinfo. Could it be that this box is not really able to do what is needed? Which flags show vt-x and vt- d?

I’m sorry, I moved on from Xen years ago, but given your processor, I would just “suck it and see”

Doing some reading, it looks like some people (me included) are having issues with the poweredge r710 and turning on virtualisation. I have a few things I can try next time I feel like taking my server down. In the mean time, things are working well enough.

Thanks for the help.

Yep, it’s all about the BIOS and the motherboard, I have no experience with the r710, but the CPU is capable, good luck . . .

An alternative is to run DAHDI on DOM0 as a “dynamic eth” span (no asterisk needed) , then on the dependent domains run DAHDI similarly (asterisk needed there to land the calls) , you can even “cross-connect” different channels to seperate dynamic spans, in effect you have “TDM over ethernet” instead of virtualizing the PCI bus, this almost always works as it is just layer 2 and that is intrinsic to Xen.

Further, you can generalize that concept to any form of Virtualization, without needing VT-d or AMD-Vi. if using AMD processors

Just in case anyone else finds themselves as confused as I did, I thought I’d post what I found out about the vmx flag. When you boot under xen, since xen is using that features, it removes the flag from the /proc/cpuinfo since it is not available for anyone else to use. In short, I have vmx properly turned on, I just needed to boot under a kernel (as opposed to dom0) iin order to see it in /proc/cpuinfo.