I have destroyed and rebuild my juju environment juju-gui the same way many times with the same script, and it has worked fine. The last two times, connecting to juju-gui server, firefox hangs spinning. It has not timed out after over 30 minutes. Apache does not get installed on the juju bootstrap/juju-gui system. A search of this site does not inform me what web server juju-gui uses, so I can delve into it deeper. I can do juju status, and it says juju-gui, and all my other charms, are happy.

In the juju-gui log it mentioned that during the compile of tornado speedups, it could not find Python.h, so it would be slow. I do not know if this happened in previous builds of the environment, as I did not keep their logs. The error says to add

sudo apt-get install build-essential python-dev

So I added into juju-gui's install script. Does anyone have any ideas on where to proceed with debugging? [edit] Adding the apt-get to the install script killed the install, as a result, I took it out. I do not recommend it.[/edit]

[Edited]Here is the script I use to rebuild my environment that has been working and then stopped.

#!/bin/bash
time juju destroy-environment -y -e maas --force
time juju bootstrap -e maas --upload-tools=true --metadata- source="~/.juju/sync-tools" --to jujuBS.local
time juju deploy --repository=".juju/charms" local:juju-gui --to 0
time juju expose juju-gui
time juju deploy --repository=".juju/charms" local:octave-controller
time juju deploy --repository=".juju/charms" local:octave --to 0
time juju add-unit -n 2 -e maas octave
time juju add-relation octave-controller octave
time juju expose octave-controller
time juju expose octave
share|improve this question
    
I think it is a node app, so no Apache needed. You expose the service? – Brandon Clark Mar 24 '15 at 22:35
    
I edited my question to include the script I use to rebuild the environment each time. – rmustakos Mar 24 '15 at 23:31
    
I understand that it is a node, but it is still serving up a web page, and something must be handling that serving. Delving deeper, it looks like it is it's own server, written in python, stored in 'charms/trusty/juju-gui/server/guiserver*.py'. Not sure why it's done this way, but I'm just a research guy trying to bring up a n HPC cluster. – rmustakos Mar 24 '15 at 23:43
    
That is an odd error message. Did you recently change the images you were using in your environment? Do the images you're using not have python installed by default? If you join #juju-gui on freenode we can help you debug and get this issue resolved. – hatch Mar 25 '15 at 1:55
    
I have made no changes to my environment. But I assumed that I was clearing the nodes out every time i did an environment destroy and rebuild. Seems that is not the case. Just got in another 4 blade server, wired it up, ran the same script on the same maas server laptop, and I am getting different results. It is not using my ntp server, even though I have it in the preseed. I am feel like it's biotch. Sounds like good time to go home and drink. – rmustakos Mar 25 '15 at 2:48
up vote 1 down vote accepted

My ubuntu server required me to update yesterday. The issues seemed to start at that point. This looks like it has reraised this bug:

https://bugs.launchpad.net/juju-gui/+bug/1397296

We tried a bunch of different things that made no difference. In the end, I switched to chrome and it worked immediately. Much thanks to @hatch!

So, something in the update either broke firefox, or broke something in juju-gui that firefox cares about. If you have not done a juju bootstrap in a while and you use firefox, if you do a bootstrap and it stops working, this chrome's for you.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.