Open Arif59 opened 3 years ago
You need to install cloudflared on your local machine.
Please try installing cloudflared on your LOCAL machine, and make sure it is in the path after you download it (ie. sudo cp ./cloudflared /usr/bin
).
You need to download cloudflared to your local PC and extract it.
Then, copy it to the directory in one of PATH
environment variable.
Following cloudflared command specified in -oProxyCommand option in SSH command print by remocolab is executed on your local PC when you run the SSH command.
-oProxyCommand="cloudflared access ssh --hostname %h"
You don't need to download cloudflared to colab instace. remocolab.setupVNC function automatically download cloudflared to colab instance.
You need to download cloudflared to your local PC and extract it. Then, copy it to the directory in one of
PATH
environment variable.
I must be doing something dumb because now I'm getting a new error:
websocket: bad handshake
ssh_exchange_identification: Connection closed by remote host
I've attached a screenshot of my local machine terminal to show that this error occurs despite the PATH
environment variable working properly.
It seems your colab instance was shutdown due to you didn't used google colab for long time. Please try again.
I have a similar issue too. Here is my error log
$ cloudflared -V
cloudflared version 2020.12.0 (built 2020-12-08-2302 UTC)
$ ssh -o UserKnownHostsFile=/dev/null -o VisualHostKey=yes -oProxyCommand="cloudflared access ssh --hostname %h" colab@lets-dairy-needed-cassette.trycloudflare.com
websocket: bad handshake
kex_exchange_identification: Connection closed by remote host
Connection closed by UNKNOWN port 65535
$
Did google block cloudflared based argotunnel too?
I tested remocolab with Argo tunnel now and it worked. Maybe Argo tunnel's server was temporary down?
I am having similar issues, I was getting "no such host kex_exchange_identification: Connection closed by remote host" using Argo tunnel so I restarted the session, and tried it again, and this time I was able to start VNC, but within a few minutes the connection disconnected and my Colab session was completely restarted. Been having this problem for the past 2 days now.
I have tested remocolab today. At first, when I try to login to the ssh server, I got following error:
dial tcp: lookup eos-century-discover-garage.trycloudflare.com: no such host
kex_exchange_identification: Connection closed by remote host
I waited few minutes and tried again but got same error.
Then, I did factory reset runtime on google colab and run remocolab.setupVNC
again.
It worked and I can access it using VNC.
I can use it at least 30 minutes.
But few month ago, I experianced google colab always disconnected after about 5 minutes. I could connect again and run some code but got disconnected in 5 minutes.
We use google colab and Argo Tunnel. They are free service and sometimes not stable.
I have tested remocolab today. At first, when I try to login to the ssh server, I got following error:
dial tcp: lookup eos-century-discover-garage.trycloudflare.com: no such host kex_exchange_identification: Connection closed by remote host
I waited few minutes and tried again but got same error. Then, I did factory reset runtime on google colab and run
remocolab.setupVNC
again. It worked and I can access it using VNC. I can use it at least 30 minutes.But few month ago, I experianced google colab always disconnected after about 5 minutes. I could connect again and run some code but got disconnected in 5 minutes.
We use google colab and Argo Tunnel. They are free service and sometimes not stable.
Thank you for testing! I tried it again today and I was able to connect and use it for around 5 minutes and then it got disconnected, when I ran a cell in colab, but this time my session didn't get wiped off like earlier, Next I checked the cloudflared.log and found this at the end:
[2021-01-13T11:38:54.283001093Z]: mux - read: read tcp 172.28.0.2:47630->198.41.200.113:7844: read: connection reset by peer: frame read error [2021-01-13T11:38:54.283185533Z]: mux - metrics: Stopping mux metrics updater [2021-01-13T11:38:54.283238093Z]: Serve tunnel error: read tcp 172.28.0.2:47630->198.41.200.113:7844: read: connection reset by peer [2021-01-13T11:40:50.787151068Z]: Initiating graceful shutdown due to signal interrupt ...
I then killed the port that was opened by cloudflared and deleted the binary files and rerun remocolab.setupVNC
and this time my connection seems to be fine, but I have not run any cells in colab, I get the feeling the tunnel would get terminated once I run a cell.
I keep getting the following error on my local terminal when I try to ssh to Google Colab using Argo tunnel:
Here's the code I ran:
!wget https://bin.equinox.io/c/VdrWdbjqyF/cloudflared-stable-linux-amd64.tgz && tar -zxvf cloudflared-stable-linux-amd64.tgz
After this extracted the cloudfare file, I ran the following:
I ran it without GPU, by the way. My colab also disconnects a minute or two after my failure to ssh. Any help is appreciated!