Cryout - please add a warning to the docs

[Michael_Davydov (Nix)]

@flamingm0e

You simply connect a VM/CT to the vmbr0. You don’t have to do anything special. A VMBR acts like a switch. It doesn’t have to go to the router.

I meant how to connect HOST to vmbr?

@flamingm0e

Edit the LXC config in /etc/pve/lxc/ and add
mp0: /location/of/host/mount/directory,mp=/mount/point/on/container

Thanks!

[flamingm0e]

@Michael_Davydov

I meant how to connect HOST to vmbr?

The host is already connected to the VMBR. It’s literally running on the host.

[Michael_Davydov (Nix)]

And it’s IP?

[flamingm0e]

It’s the IP on VMBR0, the IP you use to communicate with it.

[Michael_Davydov (Nix)]

I mean where do I set it? When setting up vmbr? You said to not put anything in there

[flamingm0e]

@Michael_Davydov

I mean where do I set it? When setting up vmbr? You said to not put anything in there

VMBR0 has an IP address already. It’s literally the one you are using to manage the system. If it didn’t have an IP you couldn’t access it.

VMBR0 should have an IP address. I told you not to add an IP to VMBR1 (the NEW vmbr you created, to act as an internal ONLY switch)

[Michael_Davydov (Nix)]

I’m talking about VMBR1 ip of host

[flamingm0e]

You should NOT have an IP on VMBR1.

You know, you should just use VMBR0, as using a secondary NIC is turning out to be more advanced than you need. You asked if it was possible and I gave you the info you need. You don’t NEED it though.

Just put every VM and CT on VMBR0 with an IP address on your network with a single NIC. Stop worrying about it.

[Michael_Davydov (Nix)]

Then traffic will go through the router, no?

Also everything is already working, I was just asking if I can access host from container through vmbr1

[flamingm0e]

@Michael_Davydov

Then traffic will go through the router, no?

NO, I keep saying this. A VMBR acts like a dumb switch. A switch never traverses a router unless it needs to ROUTE traffic.

@Michael_Davydov

I was just asking if I can access host from container through vmbr1

You don’t need to. Communication between host and VMs/CTs on VMBR0 stay on VMBR0 and never go to the router.

[Michael_Davydov (Nix)]

Ah got it thanks.

So, say, if I have VMs on 192.168.x.x network traffic between VMs wont go through the router? I think I asked that at the very start of the discussion

[Michael_Davydov (Nix)]

I think I’m almost there

One more question - I should not use mp0: /location/of/host/mount/directory,mp=/mount/point/on/container mounting to mount drive to multiple containers? How safe is it? Could I just go with containers instead of VMs and share data like that?

[flamingm0e]

@Michael_Davydov

I think I asked that at the very start of the discussion

And I answered this.

Cryout - please add a warning to the docs

Your traffic should never go to the router running VM to VM. This has nothing to do with getting it’s own IP address. There is absolutely a way to do this. It’s a little more advanced setup, but not hard. It’s simple to configure the host. Create a new vmbr (it’s essentially a virtual switch) and don’t connect the vmbr to physical adapters. Boom, you have a VM-only LAN.

@Michael_Davydov

One more question - I should not use mp0: /location/of/host/mount/directory,mp=/mount/point/on/container mounting to mount drive to multiple containers? How safe is it? Could I just go with containers instead of VMs and share data like that?

If multiple systems try to lock the same file you can have issues saving/updating a file, but if they are just reading and writing different files, it’s not a problem.

[Michael_Davydov (Nix)]

@flamingm0e

And I answered this.

Ah, I misunderstood it, I thought you meant it’s a guideline, not description of how vmbr works.

@flamingm0e

If multiple systems try to lock the same file you can have issues saving/updating a file, but if they are just reading and writing different files, it’s not a problem.

Got it. So with plex\transmission it could work, but I prefer to be safe.

[Michael_Davydov (Nix)]

@flamingm0e Phew! Thanks for all the help and patience! And thanks for not bailing out in the middle of it and suggesting to “just use OMV\unRaid\Windows\VHS”

Proxmox is up, network works with 192.x.x.x and 10.10.x.x IPs on 2 vmbrs, fileserver container succesfully serves my 2 drives over samba and NFS (read that NFS is much better and faster for linux and plex in particular), and Docker host VM is up and running. Plex is currently updating it’s library with the new locations, scanning will take a long time I assume (sigh). Wish there was some progressbar on that and not just a rotating icon.

Next step - Hass.io in another VM! Should I try to install it as HassOS or just another debian\ubuntu VM and use hassio script?

And then everything else - transmission, downloaders and stuff.

And I’ll need to find a nice way of managing all the multiple IP addresses and ports of all VMs… I was using my custom dashboard before, with a menu with links that was generated from yaml config. Thinking maybe I should just go Heimdall route or something similar. Suggestions?

EDIT: Hm , it seems that something’s wrong somewhere. Or maybe it’s just a display issue. But Proxmox shows network traffic of just about 300k (kb?) up and down, and CPU load 2-4%. RAMs growing. That’s while Plex is scanning Library, feels like there’s some bottleneck somewhere.

[Michael_Davydov (Nix)]

Ok, ended up just using another Debian VM and installing docker and using hass install script, but edited it to include the right DNS server for my LAN. Copied over more or less important contents of old /usr/share/hassio folder into the new one (i assume I also copied over some junk files but… meh). Aaand it seems to be working. Yay!

DNS is also apparently working, so I suppose the idea to reinstall it was a right one

Addons did not move over though.

EDIT:

Nginx installed in a separate CT, https://hass.domain.com is up and running again. Everything works even pushbullet now, except for some things which relied on my dashboard with custom api service, which I will replace with something.