Tiler AWS - http 502

[Filipe Marques]

I'm having an issue with the tiler map in the AWS, I always get a 502 and I already checked if the settings.json is correct. Is there any additional configuration that I need to do in the django app to make it work? The tiler app is in the /usr/local/tiler/ same machine as the django app(/usr/local/otm/app). I already tried to re-do the installation guide step 5.

Best,
Filipe

[Justin Walgran]

Filipe,

A 502 Bad Gateway error usually means that nginx is running, but the tiler is not. Do you have a tiler process running? 

$ ps aux | grep tiler

root     24339  0.0  4.9 1306192 190944 ?      Sl   Apr09   3:08 /usr/local/bin/node /usr/local/tiler/server.js

If so, it would be a good idea to check the nginx error log for evidence of why it cannot connect to the tiler upstream.

-Justin

--
You received this message because you are subscribed to the Google Groups "opentreemap-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to opentreemap-us...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Filipe Marques]

Justin,

I couldn't understand the output but it seems something isn't right on my side.

root     16259  0.0  0.1   8100   936 pts/2    S+   14:16   0:00 grep --color=auto tiler

I took a look at the nginx error log file and got this, this is only one entry. My database is named otm2_db and I checked it again just to be sure. The user and the password in the settings.json file is also correct. I tried restarting the tiler service using but got nothing. I think I'm missing some configuration script, I looked at the files that from step 6 and 7, and they are all where they should be.

2015/04/29 13:20:21 [error] 1454#0: *5350 connect() failed (111: Connection refused) while connecting to upstream, client: MY_IP, server: , request: "GET /tile/c81e728d9d4c2f636f067f89cc14862c/database/otm2_db/table/treemap_mapfeature/11/758/1161.png?instance_id=1 HTTP/1.1", upstream: "http://127.0.0.1:4000/c81e728d9d4c2f636f067f89cc14862c/database/otm2_db/table/treemap_mapfeature/11/758/1161.png?instance_id=1", host: "54.200.101.123", referrer: "http://54.200.101.123/sampa/map/"

[Filipe Marques]

Justin,

I removed the tiler from the directory and cloned it again, moved it and tried running "npm install". I got some errors in the output, i guess it is some issue with my node installation. What do you think?

p.s.: i'm attaching the output file

[Filipe Marques]

Heads up, I tried using the same nodejs version as my vagrant instance(localhost) which is v0.10.37 and got a way worse error. I have no idea how this can work in the vagrant instance since it's not able to install the necessary modules in the EC2 instance. I guess this is a "works on my machine" issue. I did the obvious thing, copied my node_modules from the vagrant machine to the EC2. As much as I'd like to know why and how it was not possible to make it work following the installation guide(probably I did something very wrong), I have to make this work asap, so this was the way.

Thanks again for being so kind and helping me out, Justin!

[Hector Castro]

Hey Filipe,

I took a look through the log file you attached earlier in this thread
and noticed the following line:

g++: internal compiler error: Killed (program cc1plus)

Generally this error message gets produced when the compilation of a
Node extension exhausts all of the available memory on the machine.
One way to avoid this is to redo this process on a machine with more
RAM, another is to enable swap and allow pages of memory to spill over
to disk (significantly slower than RAM, but should eventually
complete).

Hope that helps. Sorry for not catching it sooner.

--
Hector Castro, Operations Engineer

Azavea | 340 N 12th St, Suite 402, Philadelphia, PA
hca...@azavea.com | P 215.558.6191 | F 215.925.2663
Web: azavea.com | Blog: azavea.com/blogs | Twitter: @azavea