1

I'm developing my deep learning program on the Google Cloud Platform (GCP) and want to visualize the results using chainerui, but it's not been successful yet. Has anybody suceeded in it?

Since I login to GCP from my Windows computer at home via ssh, all the work has done by CGI basis. In the ssh window, I put a command line "chainerui server", and it runs a server on localhost (port=5000) in the GCP instance showing no problem. So, I put "http://x.x.x.x:5000" in the address bar of the Chrome browser in my home computer, however, it fails to connect to the address. (x.x.x.x represents the external address of my GCP instance)

Although I tried the --port option of chainerui to change the port to 8080 or 80, there was no change. Adding the firewall setting to my GCP instance to allow the port=5000 did not help. However, when I run the Apache2 server in the same instance and put the adress, just "x.x.x.x", in my browser, it shows successfully the Apache's test page.

I think that I need to know the way to forward the external address to the localhost:5000 that chainerui runs. Please advise. Thank you.

T.Y
  • 11
  • 2

1 Answers1

0

I found the solution myself.

On the ssh screen I put "chainerui server --host=0.0.0.0", and it worked!

Now I can see the graph chart with my browser by entering an address, "x.x.x.x:5000"

It seems that the local address, 0.0.0.0 is the same as localhost, but accessible through the external address.

I don't know if this is the best way, but it works anyway now. Thank you.

T.Y
  • 11
  • 2
  • I have to add that you might still have to add the firewall rule described in the question to allow the trafic through the port=5000. – T.Y Sep 16 '19 at 06:37