Bolt should support asynchronous communications plugins about bolt HOT 4 CLOSED

"Lack of a response to the Bolt CLI should not be considered a failure, for large systems it is quite possible that the system will output logs and results to other systems altogether"

I'd even argue that at scale a zero-response should be expected. Look at robertdavidgraham/masscan and the performance and reliability they've seen with that tool. An asynchronous operation should be considered vital as tools need to scale. And let's remember we're talking about a tool that by definition doesn't have an agent to talk with which means there will be failures at scale that shouldn't stop progression forward with successful hosts. Doing this right will also greatly improve thread pool management....

from bolt.

One idea I've talked with a few people about is building on https://github.com/puppetlabs/pxp-agent/blob/master/exe/task_wrapper.cc (as a self-contained static binary) to run a task on a remote host and then hit an HTTP endpoint when the task finishes.

Currently it writes output to a file, which could be polled for completion and output.

from bolt.

@MikaelSmith This got me thinking and it might be nice to be able to send along checksums of files that are to be executed so that the runner knows that it is executing the right thing.

Obviously not a requirement, but it might be a nice to have in the future.

from bolt.

Definitely, passing checksums is part of Orchestrator's task running. That happens at an earlier layer in pxp-agent than the code I linked to though.

from bolt.