cdayjr / dotfiles Goto Github PK

mosh doesn't work directly to Windows, so the ports should be forwarded to the wsl host instead.

According to the mosh documentation:

https://mosh.org/

Mosh will log the user in via SSH, then start a connection on a UDP port between 60000 and 61000.

Ports 60000-61000 should be forwarded, although since this is just for my own personal use, a smaller number of ports may be used instead:

https://mosh.org/

Q: What does the message "Nothing received from the server on UDP port 60003" mean?

This means that mosh was able to start mosh-server successfully on the remote machine, but the client is not able to communicate with the server. This generally means that some type of firewall is blocking the UDP packets between the client and the server. If you had to forward TCP port 22 on a NAT for SSH, then you will have to forward UDP ports as well. Mosh will use the first available UDP port, starting at 60001 and stopping at 60999. If you are only going to have a small handful of concurrent sessions on a server, then you can forward a smaller range of ports (e.g., 60000 to 60010).

https://docs.microsoft.com/en-us/windows-server/networking/technologies/netsh/netsh-interface-portproxy

This is what we use to set up the SSH port forwarding. It seems like it can only handle a single port at a time. It would be cumbersome to loop through the same action in ansible doing something like loop: "{{ range(60000,61001) | list }}" as this would result in the command running for each of the 1000 ports. Seems like it should be possible to loop over a command in a range in powershell:

https://helperbyte.com/questions/239103/to-forward-the-ports-a-range-of-ports-using-nat-in-windows-server-20012r2-using-powershell

FOR /L %I IN (10000,1,10400) DO netsh routing ip nat add portmapping Ethernet udp 0.0.0.0 %I 192.168.137.3 %I

It would be nice to have automated testing for changes to this repository.

Need to figure out the following:

• what things need to be / can be tested
• what things should be tested but cannot be

https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners#supported-runners-and-hardware-resources

Based on this, the ansible script seems somewhat impractical; it only supports Ubuntu hosts for linux, which mean the Fedora / CentOS scripts don't get ran and currently the Ubuntu block is just for a WSL host anyway.

Some additional setup will be needed to get ansible running on the Windows side of things as well.

MacOS seems promising at least.

Because I use the latest version of the yadm submodules when available, it doesn't seem like yadm submodules make sense to use.

Instead ansible should just check if theyre on the latest and update them if not. If there are problems with the latest pulled version, the ansible script can be modified to grab a specific commit instead (or just revert to using a yadm submodule for the time being).

After setting up wsl's ssh server, it seems to work fine, but when I log out of all instances of windows after a few times it disconnects. This prevents the ansible script from completing on the WSL stuff.

Here's the error I get:

kex_exchange_identification: read: Connection reset by peer
Connection reset by 192.168.1.198 port 8022

I believe this is probably due to Windows shutting down a wsl instance if the user is not logged in as I haven't ran into the issue when logged in in another window. Will need to investigate how to prevent that.