Hello
I’d like to know if some of you use XEN to virtualised servers ?
Do you also have Interworx-cp on XEN ?
Any returns would be great 
Pascal
I know Voxel uses iworx on XEN. Maybe Michael will respond with his experiences.
Chris
I can confirm that we’re using Interworx in Xen unprivileged domains with no problems. Our systems are currently running a mix of Xen 3.0.2 and 3.0.3, and it works just fine, as long as /lib/tls isn’t renamed (MySQL and some other things won’t work without it). We install a customized glibc package with the necessary tweaks to make it compatible with Xen, so there’s no compatibility slowdown with threaded software (like MySQL).
Derrik,
Can you comment on what distros you guys use as dom0/U as well if you get a sec so folks can be aware.
Thanks for posting your XEN experience
Chris
We’re running Debian sarge in our dom0s; no control panels at all in them. The domUs running Interworx have primarily been running CentOS 4.x, especially of late (since Fedora 4 has been turned over to Fedora Legacy), but I don’t see any reason FC3 or FC4 wouldn’t work equally well.
Thanks for posting your XEN experience
No problem.
Well our ideas was to use LVM2 with snapshot to create domNs based on this snapshot. So in fact the dom0s would be VDB and we would create others VDB based on this first VM
We thougth of this because in RHEL5 LVM and XEN are in the default distro. There is also a very good graphical interface virt-manager
I’m not sure if Interworx will be compatible with the next Centos5 release and I’m not sure Interworx is compatible with LVM2
Thanks for your precedent reply
Pascal
I don’t know what you mean by VDB, but that would be workable; however, you probably won’t save a lot of space using snapshots, if any, as package updates over time will replace many of the files that would be in the original copy the snapshot is based off of. You don’t necessarily have to use the same OS/distribution in dom0 as in the domUs - you could run the newer distro in dom0 and run CentOS 4 or something else in individual domUs.
I’m not sure if Interworx will be compatible with the next Centos5 release and I’m not sure Interworx is compatible with LVM2
Interworx doesn’t know or care about LVM (and can’t see it from inside a domU if you’re using it as the storage backing in dom0 anyway); also, as I mentioned, you could run CentOS 4 or something else in the domUs, which Interworx is compatible with, until such time as Interworx is certified on a newer distribution.
Tanks
VDB = Virtual lock of Device = OS
VDB that can be show to domainU are : Files, Hdd, partition local or San, Logical Volume or some QcoW images.
The snapshots in LVM are not a copy of data but are pointors and stockage block, so they don’t take places.
In fact we don’t choose to share the file space between the domain0 and domainUs like Xen does in the demo CD with UnionFS.
Exactly, we have choosen to seprate the space file between domnain0 and invite systems. To do that we have choosen to create a Logical Volume in which one we will create an invite system, then we create snapshots of this volume and present it in write mode to others Virtual Machine.
Then a command like xm create -c vmid =1 should do the job…
We did try on a RHEL5 (xen, lvm are present by default)
In fact we try to find the best way to manage our VM. Do you use only command line or a graphical interface ?
We’ll have to do a lot of others tests.
Thanks for all your comments
Pascal
Do you use an web based interface for managing your VM ?
Some of you knows these Xen web based control panel :
Or graphical panel management :
- http://virt-manager.et.redhat.com/ (the one we currently test)
Any others ?
Does Interworx will work on a solution for Xen ?
Pascal
We are running InterWorx on OpenVZ without problems, we are trying to make it work on Xen at the moment but for some reseason we got the following issieus:
-
Last line in /etc/inittab gets removed when we reboot the DomU: SV1:123456:respawn:/usr/bin/svscanboot
-
SMTP Level Scanning & SMTP Virus Scanning doesn’t want to start.
-
Load Average is >1 while there is noting hosted on the DomU
Maybe someody can help us with some tips and tweaks.
We are running Fedora Core 6 on the Dom0 and CentOS 4 on the DomU.
CPU info (Intel Xeon 5148):
[root@xen1 ~]# cat /proc/cpuinfo | grep pae
flags : fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr dca lahf_lm
flags : fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc up pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr dca lahf_lm
flags : fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc up pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr dca lahf_lm
flags : fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc up pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr dca lahf_lm
Thanks.
Hi WebXtra
Do you mean that others services start on the domU ?
There is just the smtp ?
Could you please provide the result of these commands
xm list
xm top
Thanks
Pascal
Hi Pascal,
Regarding #1; seems that it’s not being removed anymore.
Regarding #2; Socheat found that /var/qmail/control/simcontrol wasn’t created during the install, after adding this file, it was working again.
Regarding #3; found that is has something to do with this:
[root@iworx6 ~]# ps aux | grep /bin/nash
root 230 99.5 0.0 2624 388 ? Rs Mar16 4537:11 /bin/nash /init
when I kill the process, everything looks to work normaly, I don’t know yet why this process uses all CPU. Also, /bin/nash doesn’t even seem to exist. /sbin/nash does exist. So maybe a bug in my init?
[root@xen1 ~]# xm list
Name ID Mem(MiB) VCPUs State Time(s)
Domain-0 0 3861 4 r----- 5223.1
db02095a.vm 31 127 1 -b---- 30.9
iworx6.vm 29 1999 1 r----- 273513.7
iworx7.vm 37 1999 1 -b---- 358.9
xentop - 00:58:58 Xen 3.0.3-0-1.2911.6.5
4 domains: 2 running, 2 blocked, 0 paused, 0 crashed, 0 dying, 0 shutdown
Mem: 8387872k total, 8256320k used, 131552k free CPUs: 4 @ 2327MHz
NAME STATE CPU(sec) CPU(%) MEM(k) MEM(%) MAXMEM(k) MAXMEM(%) VCPUS NETS NETTX(k) NETRX(k) VBDS VBD_OO VBD_RD VBD_WR SSID
db02095a.vm --b--- 30 0.0 130884 1.6 131072 1.6 1 0 0 0 2 0 1915 1198 0
Domain-0 -----r 5223 0.3 3953844 47.1 no limit n/a 4 4 636563 368523 0 0 0 0 0
iworx6.vm -----r 273614 100.0 2047868 24.4 2048000 24.4 1 0 0 0 2 0 9530 304238 0
iworx7.vm --b--- 359 0.6 2047816 24.4 2048000 24.4 1 0 0 0 2 0 4039 25956 0
Thanks for the help.
EDIT:
Found the CPU problem, it’s a know bug.
Adding
pkill -9 -f nashto /etc/rc.local fixes the problem.
cool
so to conclude it wasn’t a Xen problem, how did you create the centos 4 image ? (how did you install centos on the domU)
Pascal
Did you use the built-in Xen support from the Fedora Core 6 installation or did you install it by yourself? Because I can’t get those jailtime.org images running on our FC6 Xen machine.
Did you compile a custom domU kernel for this or what? Maybe you could help me out a bit 
[offtopic]
Toch heerlijk, op een buitenlands forum een NL’er om hulp vragen 
[/offtopic]
[QUOTE=Sander;12264]Did you use the built-in Xen support from the Fedora Core 6 installation or did you install it by yourself? Because I can’t get those jailtime.org images running on our FC6 Xen machine.
[/QUOTE]
Hello
What is your problem ?
When you say “Because I can’t get those jailtime.org images running on our FC6 Xen machine” what is, in fact, the problem ?
Pascal
[quote=pascal;12265]Hello
What is your problem ?
When you say “Because I can’t get those jailtime.org images running on our FC6 Xen machine” what is, in fact, the problem ?
Pascal[/quote]
xm create -c /home/xen/centos.4-4.xen3.cfg
Using config file "/home/xen/centos.4-4.xen3.cfg".
Error: destroyDevice() takes exactly 3 arguments (2 given)
Config file is correct for as far as i can see. Same config with FC6 installs (through FC6 tools) works perfectly.
We want to run 4.4 without hardware virtualisation because not all of our servers support VT yet and they are still young enough to run one or more xen images.
Well I should say it different; We are distributor for HyperVM in the BeNeLux and HyperVM images are based on those from Jailtime.org.
We just use the standard FC 6 Xen kernel without modifications, but I have to say that HyperVM always tweaks some things.
If you would like to know some more about it, PM me.
[offtopic]
Tja je moet toch wat he 
[/offtopic]
Just downloaded the 3.2 trial of Xen for my new server. I’d like to make it easy to migrate web and mail servers to a 2nd server in the future using cluster manager in Iworx.
Does anyone see a problem with creating two virtual machines now under Xen on the same server and running them in cluster mode? That would be ideal, since all I’ll need to do when I get the 2nd server is move the virtual
machine over to it.