Old zerotier version isn't pinging new one

So I have several machine that uses version 1.8.4 (forked copy), and they were working fine and ping other official application machines with no issues until update 1.10.2 rolled out, now if in ZeroTier central I add a machine A (V1.8.4) and machine B (V1.10.2), and from A I can’t ping B, I’m still able to see both are online in central portal but unreachable. How to fix it
Note, Upgrade to 1.10.2 isn’t an option in the mean time.

Hello,

They should be compatible with each other. It’s hard to guess how things are getting stuck.

If you go back to 1.8.4 does it start working again?

1 Like

I think I found the issue but correct me if I’m wrong, I didn’t install/uninstall/upgrade/downgrade any side and now I can connect like before, it seems it has to do with Zerotier Central, especially with the new limit policies, I created a new account and a new network, and everything seems to work like before!

Which lead to another question, will zerotier always support “backward compatibility” per se in future -both client side and server side like central- and will always have a connection to old versions, or will it come to a point that old versions will not work anymore? I’m asking because I have some IoT with ZT in the field (middle of nowhere), and it will be more involved to upgrade by traveling there etc.

will zerotier always support “backward compatibility” per se in future -both client side and server side like central- and will always have a connection to old versions, or will it come to a point that old versions will not work anymore?

Our plan is for ZeroTier to always be backwards compatible.

1 Like

Limit policies only affect the ability to authorize new members on a network. They do not in any way stop already authorized members from communicating on a network.

I’ve got to add my voice to this one as I have been fighting this exact same issue the past few weeks.

I have been using ZT quite successfully to link a multi node Proxmox cluster together for well over a year and just a few weeks ago went to add a new node to the network.

The new node naturally had the latest copy 1.10.1-1.10.2 installed on it and would constantly drop out and become un-pingable.

I initially assumed that it was something wrong with my configuration or another networking issue, but I ultimately had to purge the new update and install 1.8.9 to get it stable.

I’m not sure if it’s a backward compatibility issue or if the 1.10.x release is bugged, but there is definitely an issue.

When the node would lose access to the network and fail to ping/be pinged I would run zerotier-cli listpeers and would clearly still see the other peers in the network, so it always confused me why it wasn’t up.

I spent a solid week trying to see if traffic was only able to go one direction by throwing UDP back and forth and disabling every firewall in my pool and nothing changed it.

ZT is definitely the issue here.

I’m not pretty hesitant to update because I don’t want to cause more issues with my cluster than the single node, so I’ve had to hold all updates.

Please let me know if there is anything I can do testing wise to try to help narrow down the issue.

Also of note, I was able to connect to the node that is on the new version by using my cellphone running the mobile ZT while it was unable to connect with the rest of the network, and I was also able to connect with it for a short while when first booting the system. After that I would always lose connection with the other nodes.

Now that the thread bumped.

@zt-joseph
Thanks, glad to know that will be the case!

@zt-grant

Limit policies only affect the ability to authorize new members on a network. They do not in any way stop already authorized members from communicating on a network.

Then I have no rational explanation why it’s solved, all what I did is creating a new account on ZT central, new network there, and all was connected (and still) with no issue, old ZT central account will authorize it, but no ping.

@Superfish1000

Also of note, I was able to connect to the node that is on the new version by using my cellphone running the mobile ZT while it was unable to connect with the rest of the network, and I was also able to connect with it for a short while when first booting the system. After that I would always lose connection with the other nodes.

That’s exactly the first thing I tried while troubleshooting the connection, and yes the phone did ping, although I think it still at 1.8.X version so probably that’s why.
I would highly suggest to create a new ZT central account, and a new network, join your machines/nodes with the different versions and see if that solves it, it did for me so worth a try.

I remember before I had an issue with the latest ZT on linux, and had to purge it and install the older version, but it was an installation error (something with ssl I can’t remember) not like this one.

I might need to try that new account like you said. I thought I had this server back up and running by purging the program and re-installing, but I just tried to access it and it’s disconnected again.

I’m not really sure what else to check. I just can’t ping over ZT and I can ping the server just fine over the normal interfaces.

Edit, I just realized that the node that is disconnected is still able to talk to one of the other nodes on my system that is running the same software version (1.8.9 vs 1.8.5/7). Not sure why that would be, but it can still ping it. I can also still reach it from my phone which registers as 1.8.9.