Search results

Reproducing is easy. Create 5 LXC CT, run a cron to stop and start each CT every minute. Within minutes you will see issue.

My issue is exactly denos is talking about, and it has no NFS involved

I don't use NFS. I use ZFS. Only local drives. I have 25 live nodes all with local ZFS drives, all 25 modes are facing issues atleast once every 2 days.

Great news. I hope we will get the new update for proxmox soon. That will end my sleepless nights.

You are lucky. I have 25 live nodes, and for last one week, I am having sleepless nights. I lost complete faith in Proxmox.

What i meant is lxc monitor is not running like other issue. I didn't mention in detail. I assumed we all knew.

service pvestatd restart This made node green, but all CT still grey.

Today we had a different issue. We terminated a CT 154, and Node went RED. 154 got deleted, Node and all other CTs pinging fine. Result of ps aux | grep 154 root@P158:~# ps aux | grep 154 root 154 0.0 0.0 0 0 ? S< Mar02 0:00 [netns] 27 5095 0.0 0.0 113276...

I can confirm by 100% that reducing ARC cache from 8GB to 1GB reduced the issue by almost 90% But you did test with 3.14.20 and issue was not there even with 32GB ARC cache right?

Yes, I took a fresh node, installed 5 CT with centos 6. Keep it in loop to stop and start each CT every minute. And issue happened in 10-20 minutes time. Keep in mind I had ZFS, and ARC cache was set at 8GB. When I changed ARC cache to 1GB, issue reduced to almost 10%, it ran for 2-4 hours...

I have only ZFS, I never tested in ext4.

You just have 2 VM, Upto 2 or 3 it is ok. SSD is needed if you need more than 3.

For windows 2012 it really matters even if you don't run anything in it. Windows 2012 takes lot of resources when it starts.

I changed ZFS cache from 8GB to 1GB. Then issue didn't happen YET on all 25 nodes for last 36 hours.

I did a modification, and for last 36 hours I don't have the node froze issue. I changed ZFS cache from 8GB to 1GB. Then issue didn't happen yet.

Another member here has a custom built kernel module. you can contact him. denos

It is a known bug in Kernel 3.13.13. We need Kernel 4.14.20+ to get this issue solved. Proxmox not yet released it.

The issue we faced was a Debian bug, not an LXC bug. First i thought it was ZFS issue, but it is not.

RAID1 with 2x1TB for Proxmox (local) and RAID1 with 1TB SSD drives for VM storage. This is what I use and i can run upto 6 or 7 KVM or 15 LXC. And it is reasonably fast.

Yes 5.1 is the version I also use. And balloonung works perfect.