Error Starting a New Elastic Instance

[AlexBussmann]

Hello,

We are trying to start our custom Elastic Instance from the Bamboo Cloud dashboard.  We are receiving the following error:

Warning:  Configuration parameter Base Url which is set to http://... does not match current URL.  This might prevent elastic agents from running.  Please validate your configuration.

Error:  Failed to start elastic instances:  Error when starting a new instance

I'm wondering how I can get more detailed logs/information about what the problem could be?

Could the warning be caused by us using an https:// URL for the Bamboo dashboard?

Thanks,

Alex

[AlexBussmann · Accepted solution]

Given that we were unable to also start stock image instances, we revisited the configuration documentation.

The following page on Generating your AWS Private Key File and Certificate File contained the following snippet:

A login key pair that you can use to log in to EC2 instances that have been started by Bamboo. The key pair is automatically generated, either the first time you use Elastic Bamboo, or if you delete the key pair. The key pair is listed as 'elasticbamboo' in your AWS console. Bamboo does not use this key pair.

It turns out that when we initially started testing on AWS we created an EC2 Key Pair with the exact same name "ElasticBamboo" and were using that to SSH into our instances.

After removing our 'ElasticBamboo' key pair Bamboo is able to generate its own and start instances.

[AlexBussmann]

Note: we see the same failures when trying to start the stock Amazon Images as well. For reference we attempted to create a custom t2.micro instance to reduce any costs on the EC2 side following the Creating a custom elastic image - Linux wiki.

[AlexBussmann]

Detailed Update: The primary problem we are currently facing is locating the Bamboo Server logs. This is required to facilitate further debugging of why the Elastic Instance will not start. The UI does not appear to provide error details or a link to the logs. The System Information page refers to an absolute path on the system that should reference the logs: /data/jirastudio/bamboo/home/logs/atlassian-bamboo.log. We have not been successful trying to combine that path with the Bamboo Servers Tomcat root URL. We did notice that there is an https://<org>.jira.com/logs URL that is active titled "Studio Logs" but we do not seem to have the credentials required to access this. We also tried to investigate the EC2 side further to see what we could gleam. We can visually see that the Bamboo Server is creating two Security Groups called "elasticBamboo" and "ControlledByBamboo." We do not see an instance started. The Bamboo Access Key ID/Secrets have been associated with Administrator privileges for the time being to remove the possibility we are seeing further permissions issues. Since we cannot rule out that the instance is starting and shutting down immediately we dug a bit into trying to run the Elastic Image manually from the AWS dashboard. This requires a bit of trickery and is ultimately a dead end. For reference we needed to: - comment out the "telinit 0 0" calls in /opt/bamboo-elastic-agent/etc/rc.local which were being called when the startup detects you are not launched by a Bamboo Server - touch /root/.startedFromBamboo - create a new AMI from these change At this point you can SSH into and attempt to validate the custom rc.local script. Unfortunately this fails because, we assume, that Bamboo specifies some custom Instance Metadata and User Data setup that isn't accessible to the instance (ie: we are not launched from the Bamboo Server). These failures end up written out to /home/bamboo/bamboo-elastic-agent.out. From what we can tell the EC2 side image looks fine. At this point we feel we've reached a dead end in our ability to investigate this further.