Local galaxy - find broken tools

, ,
#1

Hello admins,

we have a problem with a new Galaxy version after trying to install all ~780 tools from the big public servers, using ephemeris.

Most tools are installed successfully, apart from 2 or 3 (R, scikit etc)

It looks like the var options is not OK (removed here for clarity) and config.set cannot be set properly (red underlined in Chrome browser developer tools).

My question :
is there a way (without the Galaxy GUI) to
find broken tools,
or remove just broken or incomplete tools ?
Or should we reinstall all tools ?
Or is there a subset of “easy to install tools” ?

Version Galaxy 2005, installed with GalaxyKickStart, on Ubuntu 20.04

Thanks!

// js-app.mako
    var options = {"root": "/", "user   
  // Removed for clarity
"stored_workflow_menu_entries": []}, "params": {}, "session_csrf_token": "db80f75b44e70cee"};
    var bootstrapped = {};

config.set({
    options: options,
    bootstrapped: bootstrapped,
    form_input_auto_focus: false
});
1 Like
#2

How do you mean broken tools? In what way are they broken?

#3

They are broken in the sense that they do not install properly - conda is still running on the VM after 24-48 hours. Therefore we kill conda.

After that, and following restarting galaxy, the entire server is not accessible as before in the browser. Supervisor reports all processes to be running. We think it is due to a failed tool installation, which is probably present in this line.

var options = {"root": "/", "user

That is what is reported by chrome dev tools.

But we aren’t sure.

How can we check if tools are installed properly without a GUI ?

#4

It isn’t clear in the code you are pasting what it is referring to. That also appears to be javascript, while Galaxy is written in python.

You can directly query the galaxy database to check on the state of things if the galaxy daemon isn’t starting. Otherwise you can check the output of the API. I don’t think I have ever encountered an issue with a failed tool installation bringing down galaxy. Are you sure you do not have two issues?

1 Like
#5

The javascript code is literally what I get in chrome devtools when trying to access my local Galaxy server (it doesn’t respond). Of course Galaxy has a lot of the front end written in Javascript.

Thanks for the idea about the API, but this might be tricky with an unresponsive server (I’ll try it). Any ideas where I can find docs on directly checking for errors on the Postgres DB backend ?

It’s good to know that you haven’t seen tool installations bringing down Galaxy. I see on this GAT2021 - Day 1 - Installing Galaxy with Ansible - YouTube that usegalaxy uses a different DB for tools to the main DB, but that’s not a common approach.

We might well have several issues, we’re aware of that. But it’s nice to be able to learn about Galaxy by breaking and repairing it before we can it and start again (its GalaxyKickStart, so not too tricky).

Thanks

#6

Your first starting point should probably be the Galaxy log, it does an ok job of reporting failures.

1 Like