I think I traced it to some Docker Containers. I'm not able to use Plex or Pihole without a crash in 3 days. I'd really love to get this solved. I've realized that I can run the following Docker containers without error:

- DuckDNS

- mariaDB

- Swag

- Nextcloud

If I run Pihole or Plex I get a call trace or kernel panic within 3 days. Does this have something to do with br0? This happened last night during an appdata backup and may have stopped the backup. Can anyone translate these errors to human readable?

 

tower-diagnostics-20210204-1128.zip

 

Edited February 4, 20215 yr by adminmat

18 minutes ago, adminmat said:

Does this have something to do with br0?

Most likely yes.

google macvlan crash site:forums.unraid.net

1 hour ago, adminmat said:

Does this have something to do with br0?

 

I see you posted in that thread a couple of weeks ago.  Sorry, I missed it. 

 

Yes, the thread you linked about setting up VLANs is the one I used to create a br0.3 network for docker containers and it is still applicable.  This eliminated my macvlan/broadcast call traces.  YMMV.

Edited February 4, 20215 yr by Hoopster

1 hour ago, Hoopster said:

I see you posted in that thread a couple of weeks ago.  Sorry, I missed it. 

 

Yes, the thread you linked about setting up VLANs is the one I used to create a br0.3 network for docker containers and it is still applicable.  This eliminated my macvlan/broadcast call traces.  YMMV.

 

Thanks for the info @Hoopster. Probably a dumb question (and I'm about to read through the thread again) but do I set up the VLAN in unRAID? Or on my router? I, like you will be using one interface / ethernet port on my motherboard.

Edited February 4, 20215 yr by adminmat

1 hour ago, adminmat said:

 
Thanks for the info @Hoopster. Probably a dumb question (and I'm about to read through the thread again) but do I set up the VLAN in unRAID? Or on my router? I, like you will be using one interface / ethernet port on my motherboard.

Some setup is required on both the router and in UnRAID.


Sent from my iPhone using Tapatalk

Edited February 4, 20215 yr by Hoopster

So I set up the networks per the guide. I went to start VM service and got an error that the libvert service failed to start. I figured I'd try to reboot. When I went to reboot it would not let me stop the array. I then clicked shut down (which I assume that stops the array before it shuts down) It then hung and became unresponsive.

 

After reboot it now says unclean shutdown detected. And started a parity check. OMG. I just had one 3 days ago.

 

Here are a couple screenshots from my KVM console when it shut down:     Do I have to let this finish?

 

 

 

This is where it hung:

 

I was able to set up the VM manager so my router is providing a DHCP lease for my linux mint VM. However, I'm not able to configure my Docker settings as the guide suggests. Bonienl suggests you set up a VLAN network for Docker then says that your router will not know that this network exists. Then he says to choose a IP range that is out of the range of your Docker settings.

 

I set it up as he did but in his example he is setting his IP address assignments in Docker network setting to automatic. But he is using the IP of the netowrk mentioned at the beginning of the write up. This is contradicting.

 

When I try to open Plex for example I does not find it. Is having segregated VLANs really the only way to get docker containers to work on unRAID?

Edited February 4, 20215 yr by adminmat

Ok @Hoopster, little update. I was able to get my docker containers working again. Set up br0.50 for VMs and br0.60 for Docker. Added the VLANs on my Edgerouter. Tagged the ports on the switch, updated each container and I'm up and running. Couple question:

 

- The way I understand it, my router wont be handing out DHCP leases for Docker containers like it does for VMs? (I'm so confused how Docker works)

- Do I have to completely segregate the VLANs with firewall rules to keep the Call Trace issue from happening?

- Should I delete/disable br0 completely?

- I have "Host access to custom networks" set to "Enable." I think I need this for Nextcloud to work (?) Will this cause the macvlan issues?

Edited February 4, 20215 yr by adminmat

2 hours ago, adminmat said:

Is having segregated VLANs really the only way to get docker containers to work on unRAID?

No, but, if you are assigning custom IP addresses to docker containers and you are getting macvlan call traces, it is one way to prevent those in many cases.

 

2 hours ago, adminmat said:

- The way I understand it, my router wont be handing out DHCP leases for Docker containers like it does for VMs? (I'm so confused how Docker works)

Docker has no idea about the existence of your router based DHCP server and does not communicate with it.  If you want to do DHCP with docker containers, the VLAN from the unRAID/docker side needs it own IP address range.  I don't even deal with that and just assign static IP address to docker containers within the range I told it to use which is outside the range of DHCP addresses I let my router assign (so there are no conflicts).

 

2 hours ago, adminmat said:

- Do I have to completely segregate the VLANs with firewall rules to keep the Call Trace issue from happening?

Not necessary.  I also have a Ubiquiti USG router (the one used in the bonienl example) and by default it allows all corporate LANs (my VLANs are in this category) to communicate back and forth.  I have done nothing to prevent this with firewall rules.

 

2 hours ago, adminmat said:

Should I delete/disable br0 completely?

Again, not necessary.  It's mere presence (at least from what I have seen) does not cause any problems, just assigning IP addresses on this interface to docker containers is the issue.  I still have br0 enabled.

 

2 hours ago, adminmat said:

- I have "Host access to custom networks" set to "Enable." I think I need this for Nextcloud to work (?) Will this cause the macvlan issues?

I don't know if this, by itself, could cause the issue.  I had the call trace issues long before that option even existed.  I do currently have it disabled and have never enabled that option.  There are some reports of networking/router issue with it enabled, but, those just may be DHCP conflicts.

22 minutes ago, Hoopster said:

 

Quote

Docker has no idea about the existence of your router based DHCP server and does not communicate with it.  If you want to do DHCP with docker containers, the VLAN from the unRAID/docker side needs it own IP address range.  I don't even deal with that and just assign static IP address to docker containers within the range I told it to use which is outside the range of DHCP addresses I let my router assign (so there are no conflicts).

Got it. I have the Docker containers set to static IPs within VLAN60. And I just limited my router's IP range for VLAN60. Which I now understand as I just had to add my HDHomerun to that network and I can see how it could create a conflict.

 

22 minutes ago, Hoopster said:

 

Quote

I don't know if this, by itself, could cause the issue.  I had the call trace issues long before that option even existed.  I do currently have it disabled and have never enabled that option.  There are some reports of networking/router issue with it enabled, but, those just may be DHCP conflicts.

I have since disabled it. I can't remember the reason I needed it now.

 

This is great. I'm glad I got this sorted. Fingers crossed that I don't have any more issues. 🤞 You don't have to have an HDHomerun that you are using across the VLANs?
 

Thanks for all the help

Edited February 5, 20215 yr by adminmat

9 minutes ago, adminmat said:

You don't have to have an HDHomerun that you are using across the VLANs?

I have two HDHomeRun Connect dual-tuner boxes.  They are integrated with Plex for Live TV and DVR.  Currently, they are on the same subnet as unRAID server.   In my network setup with communication between 'corporate' LANs enabled by default, it may still work if I moved them to a VLAN.  If have just never had a reason to do so.