25.03-BETA on hyper-v 100% CPU-usage
-
I tried it for a short time on hyper-v. It became very slow after some time, dashboard showed 100% CPU-usage, hyper-V only showed a very small CPU-usage. So I reverted back.
Side note:
Because of other problems in the past, SCSI errors, I installed pfSense 24.11 fresh, again with ZFS but this time in a fixed vhdx-file instead of dynamic. I hope that will fix that problem. If not, I have to go with UFS in the future. I already changed the underlying SSD but the problem was persistent and is not related to the actual storage. -
@Bob-Dig said in 25.03-BETA on hyper-v 100% CPU-usage:
If not, I have to go with ext4 in the future.
Sure about that? I'll first try UFS ;)
-
-
J jimp moved this topic from Development on
-
Any more info on what was using the CPU during that time? You can see more info on the System Activity page.
-
@marcosm said in 25.03-BETA on hyper-v 100% CPU-usage:
Any more info on what was using the CPU during that time?
No, I didn't see anything that popped into my eyes, maybe something with php...
But the day before, on 24.11, I noticed it too, that pfSense was very slow. In hyper-V I saw the SCSI-errors in the console again. And looking at the System Activity page, it was trying to compress some DHCP or some leases data(?), for hours.
-
@Bob-Dig That sounds like kea2unbound which is used for DHCP DNS registration in Unbound. I believe @cmcdonald is improving its efficiency.
-
@marcosm Today I noticed the SCSI Error again (24.11), looked in the logs for it and could track it down to when Veeam was doing its backups of my hyper-v VMs like pfSense. So I guess, it is somewhat to be expected.
Feb 12 07:51:16 kernel (da0:storvsc0:0:0:0): Retrying command (per sense data) Feb 12 07:51:16 kernel (da0:storvsc0:0:0:0): SCSI sense: UNIT ATTENTION asc:3f,2 (Changed operating definition) Feb 12 07:51:16 kernel (da0:storvsc0:0:0:0): SCSI status: Check Condition Feb 12 07:51:16 kernel (da0:storvsc0:0:0:0): CAM status: SCSI Status Error Feb 12 07:51:16 kernel (da0:storvsc0:0:0:0): WRITE(10). CDB: 2a 00 00 9e 90 88 00 00 08 00 Feb 12 07:57:36 kernel (da0:storvsc0:0:0:0): Retrying command (per sense data) Feb 12 07:57:36 kernel (da0:storvsc0:0:0:0): SCSI sense: UNIT ATTENTION asc:3f,2 (Changed operating definition) Feb 12 07:57:36 kernel (da0:storvsc0:0:0:0): SCSI status: Check Condition Feb 12 07:57:36 kernel (da0:storvsc0:0:0:0): CAM status: SCSI Status Error Feb 12 07:57:36 kernel (da0:storvsc0:0:0:0): UNMAP. CDB: 42 00 00 00 00 00 00 00 18 00 -
@Bob-Dig said in 25.03-BETA on hyper-v 100% CPU-usage:
could track it down to when Veeam was doing its backups of my hyper-v VMs
How is that configured? Backing up the VM disks in hyper-v or some client runnig in pfSense backing up directly?
-
@stephenw10 said in 25.03-BETA on hyper-v 100% CPU-usage:
Backing up the VM disks in hyper-v
This! And I am using only the free Veeam Agent for Windows.
Today I noticed the SCSI Errors again. And looking in the logs, again at "backup"-time. Wonder why I didn't saw that before. Most probably because I often look in the firewall log first, then deleting the whole log. When I then later in the day saw the SCSI Errors in the console, I never could track down the time, when this might had happened.
But most of the times, I don't have any problem with these SCSI errors. Might just be a coincidence that one time this compressing of lease data or whatever this was, was happening at the same time. Or maybe it wasn't happening at the same time and it had another problem (making 100% CPU usage and not stopping), not related to the SCSI Errors at all. And all of that happens/d on 24.11.
The problem with 25.03 was different, maybe related to the dashboard, because there, hyper-v wasn't showing the high CPU usage, only the widget showed it. But I have no "needs" to run the beta again. My hyper-v pfSense is my main and only router, it has to run.
-
Hmm. If you boot a test pfSense VM in Hyper-V does it also show that?
I'm not sure if any testing in hyper-v is happening otherwise.
-
@stephenw10 said in 25.03-BETA on hyper-v 100% CPU-usage:
If you boot a test pfSense VM in Hyper-V does it also show that?
You mean the 25.03-BETA problem? Even if I would set one up, I "lost" all my registrations for another Plus. But even if I could do it, it wouldn't be in much use. Really the only thing I can do is a small test run on my main pfSense but I will need to revert back after some hours.
I am already thinking to move my pfSense-VM into the nested Proxmox host I have running in Hyper-V. So much fun with all the switching/bridging. And I would need another activation for sure. In the end, I am just an (ab)user
