Every virtual machine I restart doesn't boot.

[Pilow]

@ohthisis I see nothing outstanding
you have two SRs, thick provisionned
one small 95Gb that is empty probably created on the install, and one big 3.64Tb on /dev/sdb

VMs are on the big SR, same sized VDIs could indicate existing snapshots.

is your /dev/sdb a RAID5 array or a standlone disk ?

Can you create a new test VM, that is running normally on this SR ?

[ohthisis]

@Pilow
I guess my server using RAID5.
I created a VM with PXE as boot, but it is VM.start: 50%.

[Pilow]

@ohthisis I really don't understand what is happening to your server
all basic tests show something is wrong but I don't know what

hope that someone else could hop in on this case
@Danp ?

[ohthisis]

I rebooted the server and now I got:

[Pilow]

@ohthisis you have no bootable disk on this VM, or boot order is wrong.
could you screen the DISKS tab and the ADVANCED tab ?

[Danp]

It is difficult to know the source of the problem without more details from your logs. Can you run the following commands on your XCP-ng host and then attach the generated files here?

grep -A 15 -B 15 -i error /var/log/xensource.log/xensource > xensource.txt
grep -A 15 -B 15 -i exception /var/log/SMlog > smlog.txt
dmesg -T|grep -Eiv 'guest|capacity|promiscuous' > dmesg.txt

[ohthisis]

Hello,
This issue was related to XPC-ng and caused me to delete and recreate and configure one of the most important virtual machines I have. As I said, every virtual machine I create is only 33% complete, and when I try to turn on virtual machines that are already turned off, I get the message "VM state is halted but should be running".
I had to turn the server off and back on to fix the problem. To be honest, I'm afraid to test again to see if the problem is fixed, because I have a VPN server that has a number of users connected to it, and I also have a monitoring server.
If it is a hardware problem, the green lights on the hard drive should change to orange or red, but all the hard drive lights on the server are green. I think the problem is related to the stability of XCP-ng.

[olivierlambert]

XCP-ng is rock stable. We have (and Citrix does too) many many automated test for such basic operations. So you clearly have an issue somewhere triggering that problem.

[ohthisis]

@olivierlambert Where? How to find it? As you can see, I only have 4 virtual machines on the XCP-ng host.

[olivierlambert]

Starting with the documentation is a good thing: https://docs.xcp-ng.org/troubleshooting/

This will give you (and us) hints on what's going on