Hi, what is the safest and best way to downgrade?

 

I searched in wiki but didn't find anything.

 

TY

Tools - Upgrade OS, and revert to 6.8.3 if it's an option.  You will need to reassign your cache pool though after this is done.

 

If it's not an option to revert, then download the 6.8.3 zip file from https://unraid.net/download and overwrite all of the bz* files on the flash drive with those in the archive and reboot (You'll still need to reassign the cache pool)

 

Why do you want to revert?

Because I have problem with dockers and VM.

They absorb much more cpu that stays at 100% all the time. I hoped that 6.9.1 would solve this but nothing.

I rebuild some docker and the only vm but nothing. Moreover the vm needs more time to startup.


Inviato dal mio iPhone utilizzando Tapatalk

If I have a backup of the flash drive there’s some shortcut?


Inviato dal mio iPhone utilizzando Tapatalk

Copy the entire contents of the backup onto the flash.

I have the same issue, everything was great in 6.8.3 but now (6.9.1) when I start my Win 10 VM the CPU maxes out at 100% and stay there until I force stop the VM. 

I did make a backup before I upgraded so can I just overwrite my flash with the backup files?

Should I move all data off the Cache before I revert? It was already formatted btrfs.

crimson-diagnostics-20210321-0046.zip

On another post s guy told me that in my case could be related to my SAMSUNG SSD cache drive. I had to reformat it but I lost docker.img and libvirt.img so I’m still trying to fix everything (so annoying!).

Just for curiosity do you have a ssd evo 860 too?


Inviato dal mio iPhone utilizzando Tapatalk

On 3/21/2021 at 7:14 AM, Jokerigno said:

On another post s guy told me that in my case could be related to my SAMSUNG SSD cache drive. I had to reformat it but I lost docker.img and libvirt.img so I’m still trying to fix everything (so annoying!).

Just for curiosity do you have a ssd evo 860 too?


Inviato dal mio iPhone utilizzando Tapatalk

No, I'm running the VM on a NVME

So I downgraded to 6.8.3 and the VM problem remained, in the downgrade process I lost my Nvidia driver so I tried running the VM with VNC.  It was still slow and sluggish. So I upgraded back to 6.9.1, added the Nvidia driver again still the same problem with VM.  I even deleted the VM and started fresh (No change).  Now to add to the frustration my server keeps crashing.  I'm only running dockers, no VM's running and it is crashing. The last crash it was pretty much at idle, with one Plex direct stream.  I could really use some help.  

crimson-diagnostics-20210331-2238.zip

Edited April 1, 20215 yr by Crimson Unraider
Add more info

On 3/21/2021 at 7:14 AM, Jokerigno said:

I lost docker.img

docker.img is just the executables of your containers, those can be downloaded again, so it is easily recreated simply by creating your containers again. Previous Apps feature on the Apps page will reinstall your containers exactly as they were using the templates saved on flash.

Turl, I didn't lose my docker and libvrt images like Joker. When I reverted I copied the entire backup like Squid suggested and I was able to recover all my dockers and Plugins with the exception of the Nvidia plugin.  It had an error saying it was depreciated.  Everything I read said it was no longer supported below 6.9 so I upgraded back to 6.9.1.  Should I delete my Docker and Libvrt images and start fresh with 6.9.1?    

45 minutes ago, trurl said:

docker.img is just the executables of your containers, those can be downloaded again, so it is easily recreated simply by creating your containers again. Previous Apps feature on the Apps page will reinstall your containers exactly as they were using the templates saved on flash.

 

I restored it. But I'm still struggling to have the same performance as before.

@Jokerigno and @Crimson Unraider it sounds like you are having the same problem as me, with VM's that used to run perfect in 6.8 but now are super laggy with 6.9. 

I have not figured out a solution or even a cause, so I ended up here after I decided to "bite the bullet" and downgrade back to 6.8. 

 

 

I doubt this will help, but here is another thread where I have posted some comments discussing the issues I'm seeing. There has been some back-and-forth about similar issues, but no universal solution it seems. 

 

I finally found my issue. It was power save cpu setting. After changed to on demand everything started working again as expected.

My downgrade process has hit a bit of a snag many times along the way, so I will leave the following information here in case of other people who try to downgrade while using a very non-stock configuration

 

At first, I was hesitant to start a downgrade because I was worried it would be a lot of troubleshooting along the way, 

but then I saw this thread and it sounded like it would be simple: Just download the zip of the version I want, copy over the `bz*` files, reboot, and reassign the cache drive. 

In hindsight, I guess I should have listen to my instincts telling me that it would turn into hours of troubleshooting, because that is exactly what happened.

 

However, I will admit that most of these problems are my fault to over-complicating my Unraid setup.

 

I have a pfSense as my router, and the VM domains are on the cache drive, so...

• First, I had to find a way to reassign the cache drive... but since the pfSense VM was on the cache drive, that means the router VM did not start, so I had no way to access Unraid over the network. I had to hookup my keyboard and monitor directly and reboot Unraid into GUI mode, but at least I was prepared for that hardware-wise.
• Next, I was finally inside the Windows VM and thought everything was all set, until I realized that the VM had no internet and was failing to get an IP from DHCP. So I went poking around in Unraid again...
• I saw that pfSense was running, but all the other linux VMs had failed to start. I realized that the other VMs failed to start because they were trying to use newer versions of Q35 or i440fx that were available in unraid 6.9 but not available in 6.8, so I had to fix all those
• And THEN, with the pfSense VM still running, Unraid still had no internet because I forgot to re-add the 'vfio-pci.ids=...' stubbing to the syslinux.cfg, so Unraid was trying to use the Ethernet NIC that was being passed-through to pfSense! Somehow it was able to successfully passthrough that network card, because pfSense was working fine (because the Wi-Fi network could now access the internet!), but at the same time the Unraid OS still thought it was using that same interface, so Unraid and all the other VMs had no internet. After fixing the syslinux.cfg, I had to reboot twice for some reason before Unraid finally booted up with the motherboard ethernet as `eth0` rather than trying to hold onto the pfSense network card.

Now, it seems like I have the array and the VMs up and running correctly in 6.8, and I will now try re-enabling the dockers (I disabled the dockers because I kept getting hung-up on things due to either the cache drive being missing, or the network missing, or whatever other things happened while I was troubleshooting)

 

So I guess TLDR here is that be careful about trying a downgrade if your Unraid setup is very non-standard or very customized. Make sure to prepare to re-do any workarounds that you had previous used before Unraid 6.9 (such as the new improvements to vfio-pci binding), otherwise you will have to fight your way into the server to setup all those things after the downgrade 

14 minutes ago, Jokerigno said:

I finally found my issue. It was power save cpu setting. After changed to on demand everything started working again as expected.

I have seen this several places, and unfortunately it made no difference in my case. I was already using the "performance" setting before the upgrade, and for troubleshooting sake I also tried all the other options for that and I did not change my problem,

and I also tried changing the overclock in the BIOS, and resetting everything back to stock speeds in the BIOS, I tried all sorts of variations. None of it made any difference, the problem still persists (in 6.9)

 

Since I haven't finished the downgrade yet, I do not know yet whether downgrading to 6.8 made the problem go away, but the first time the Windows VM booted up it was running way better than before, even just the login screen during startup was way more responsive than it was before, so I think the downgrade probably made a difference

I'm also experiencing high, constant CPU usage from 6.9.2 only when a VM is booted, which is causing it to run sluggishly slow. 

I attempted to downgrade to 6.8.3 from the Unraid Boot media and transfer my config folder over. However, I can't seem to get a ping response from the gateway using 6.8.3. 

I've also ensured I have manually set the network settings within the network.cfg file with no luck. 

However if I transfer the brz* files from any 6.9.* version, the system is then pingable and I'm able to login to the console.

 

Any ideas? 

 

 

The original issue I'm experiencing is found in this thread. 

 

 

Thanks!

Edited October 11, 20214 yr by fearlessknight
Additonal Info added

On 3/14/2021 at 4:37 PM, Squid said:

Tools - Upgrade OS, and revert to 6.8.3 if it's an option.  You will need to reassign your cache pool though after this is done.

 

If it's not an option to revert, then download the 6.8.3 zip file from https://unraid.net/download and overwrite all of the bz* files on the flash drive with those in the archive and reboot (You'll still need to reassign the cache pool)

 

Why do you want to revert?

Thanks for this. Why do you need to reassign cache pool?

3 hours ago, cinereus said:

Thanks for this. Why do you need to reassign cache pool?

There was a change in Pools management in 6.9 that was not present in 6.8 and require the disks to be assigned in case of roll back to a version prior to 6.9.

2 hours ago, ChatNoir said:

There was a change in Pools management in 6.9 that was not present in 6.8 and require the disks to be assigned in case of roll back to a version prior to 6.9.

Thanks. So I will be okay going from 6.11.0 to 6.9.x?