-
Notifications
You must be signed in to change notification settings - Fork 42
device or resource busy when attempting to load elastic-agent config in docker #981
Comments
/cc @jalvz I'm hitting this as well in my tests, investigating to see what's going on |
From my research, anything that modfies the inode will fail after docker 1.10 so we may need to look into attempting to mount a parent directory. Or don't use volumes and just copy in the required configuration files within the step and cli provisioning |
Thanks for digging @adam-stokes !
What would that be? Or in other words, why this wasn't a problem before #978? |
I'm not sure why this wasn't a problem before, I think this is a pretty good description of what's happening when |
As we discussed online, maybe we do not need the volume anymore, as we are already consuming the configuration file in the latest state:
I propose removing that fetch part, as it is duplicating efforts. Therefore, we could remove the volume to the config file in the agent compose file. |
BTW, maybe @michalpristas knows about this: is the config file present/bundled into the installers (tar, rpm, deb, docker)? When we extract a TAR file:
We do see [2021-04-06T14:12:29.471Z] elastic-agent-8.0.0-SNAPSHOT-linux-x86_64/elastic-agent.reference.yml. Will that file be used by the agent? |
AFAIK, no, that file is not used. |
This should be fixed with #983 |
Could it be the case an agent with fleet-server performed an operation on the file that the simple enrollment of the agent was not performing? |
In some instances like stand-alone testing and running from the cli we hit this error:
If we remove that volume mount everything works as intended, however, this makes it a little difficult to include the default system integration
The text was updated successfully, but these errors were encountered: