DockerVM runs but access to console fails

[-fun-]

So I got my DockerVM / RancherUI running. Next step according to the guide would be to connect to the VM using cu. The UI provides "State: RUNNING Com Port: /dev/nmdm1B". So I execute:

cu -l /dev/nmdm1B

As a result I get a login prompt to a completely different VM!

After iohyve came to FreeNAS - without UI support back then - I created 2 VMs. One is a template based on an ArchLinux distribution and the other is a tvheadend VM based on this template. The above cu command provides the login prompt to the tvheadend VM. Originally I would connect to this VM via ssh or using iohyve console tvheadend.

How can I connect to the new DockerVM?

In /dev I can see nmdm1A and nmdm1B. Neither works.

Can the existing VMs coexist at all with the new stuff? Or should I remove the old VMs from the system and if so, how would this be done?

 

[Yusuf Limalia]

So I got my DockerVM / RancherUI running. Next step according to the guide would be to connect to the VM using cu. The UI provides "State: RUNNING Com Port: /dev/nmdm1B". So I execute:

cu -l /dev/nmdm1B

As a result I get a login prompt to a completely different VM!

After iohyve came to FreeNAS - without UI support back then - I created 2 VMs. One is a template based on an ArchLinux distribution and the other is a tvheadend VM based on this template. The above cu command provides the login prompt to the tvheadend VM. Originally I would connect to this VM via ssh or using iohyve console tvheadend.

How can I connect to the new DockerVM?

In /dev I can see nmdm1A and nmdm1B. Neither works.

Can the existing VMs coexist at all with the new stuff? Or should I remove the old VMs from the system and if so, how would this be done?

Silly Question by can you send a Screenshot of your VM page?

 

[-fun-]

Sure. Not a silly question. it's rather a silly problem.

My VM:

These are the attempts to connect to the VM. Note the RancherUI VM is not listed with iohyve list, and how all attempts result in connection to the tvheadend VM:

After stopping the tvheadend VM I still cannot connect to the RancherUI VM. Very weird.

Btw. I noticed I can't stop the RancherUI VM. Just doesn't work.

 

[Yusuf Limalia]

Okay last one.

VM's > Click on your VM > Devices > RAW > Edit

you got a similar path?

 

[-fun-]

yes, exactly the same.

 

[KrisBee]

1. Why would DockerVM (rancheros based) be in the iohyve list? It won't be, because your DockerVm was not created/started using iohyve.

2. I'd suggest stopping all VMs. Check iohyve VM properties to see what consoles they are using as there could be a clash.


iohyve conlist


Clear iohyve consoles if necessary.


iohyve conreset


Doublecheck:


ls -lah /dev/nmdm*


3. Now start DockerVm only and ssh into your FreeNAS box and check using htop or ps if you see a "grub-bhyve" process running with high CPU usage . If so, then your DockerVM is not setup correctly. Otherwise, your DockerVM may actually run correctly and try to connect to it via console. ( You can also ssh to your DockerVm at this stage if you know its IP - ssh rancher@<ip-address> )

4. Change iohyve VM properties to use non-clashing consoles. Re-start iohyve VMs.

 

[-fun-]

1. Ok, understood

2. Done / stopped. iohyve conreset did not work but iohyve set <name> con= worked. No /dev/nmdm* remaining.

3. Actually there was a CPU hogging grub-bhyve process. I killed that one and recreated the DockerVM from scratch. After starting this an image download was NOT reported / required. But there was again a 100% CPU grub-bhyve process. Connecting to this VM was not possible.

So according to your post there must be a configuration error in the VM? I checked this over and over again. Can there possibly be some other problem outside of the VM configuration?

 

[KrisBee]

I would check the rancher img has downloaded correctly, you should find it under a hidden .bhvye folder in your pool.

One common misunderstanding is that the path of the DockerVM raw file is a pre-existing dataset mountpoint + name of an image file that does NOT pre-exist, e.g:

/mnt/myPool/DockerVm/anything.img

 

[-fun-]

Instead of checking the file (how could I do that?): Can I just delete this image file and start from scratch?

I configured /mnt/flospool/vm-storage/rancherui.img, where /mnt/flospool/vm-storage is a dataset.

 

[-fun-]

... how could I do that? ...

Got it, uncompress failed.

 

[-fun-]

Got it, uncompress failed.

So the download consistently failed. (This may be a problem in FreeNAS.) I retrieved this file with wget in few seconds without any problems into the required folder and ...

... it works! I'm seeing the RancherOS Logo right now.

Thank you!

 

[Yusuf Limalia]

Glad you got it sorted!

@KrisBee nice one!

 

[kernalzero]

3. Actually there was a CPU hogging grub-bhyve process. I killed that one and recreated the DockerVM from scratch. After starting this an image download was NOT reported / required. But there was again a 100% CPU grub-bhyve process. Connecting to this VM was not possible.

I'm having a similar issue. how did you get to your resolution?

 

[-fun-]

I'm having a similar issue. how did you get to your resolution?

The problem was caused by an incomplete download. Once this download worked (I did this manually, see above) this worked.

However independently from this problem I experienced occasional (2) situations where the VM consumed all virtual memory and blocked FreeNAS completely after all swap space was used up. I never solved that one.