Add non-blocking deployment script

[Amdrel]

Description

This patch adds a script as described in my comment in T410730 that can assist with deployment. I've left out GitHub Actions integration for now, but I can add that later if we think it's a good idea.

@don-vip I haven't tested this script against the encoders. However, I did test most of it against another machine on my network with the exception of lines 50-58 of deploy.sh. If you're okay with me testing it or you want to do it, please let me know. If we can test this against a real encoder instance I would feel more comfortable taking this PR out of draft status.

Update: I've tested this against encoding01 and encoding02 after making some fixes and it appears to be working the way I expect.

Configuration

This script relies heavily on environment variables since that would make it easier to work with in a CI/CD workflow if/when we implement something like that.

• V2C_USERNAME: The ssh username to use when connecting to the encoder
• V2C_CONSUMER_KEY: Value that gets substituted into the puppet manifest
• V2C_CONSUMER_SECRET: Value that gets substituted into the puppet manifest
• V2C_REDIS_PW: Value that gets substituted into the puppet manifest

Changes

• Add new deploy.sh Bash script to assist with deployments

[don-vip]

Hi Amdrel, sure go ahead for testing :)

[Amdrel]

I tested the script against encoder01 and I need to make a small fix to the systemd unit file that we generate. The ExecReload command in there currently is broken due to a missing environment variable resulting in options not being passed in properly. I'll have to do that tomorrow as it's getting late.

[Amdrel]

I've modified the manifest to always restart the service when applied rather than doing it manually by adding notify to the git pull and pip install steps (which always notify). I tried to eliminate any possibility of puppet doing an implicit restart of the service during manifest application without success, so an apply and manual restart as I was doing before could lead to the service being restarted twice. I decided to let Puppet handle it instead with a custom restart command since it's intelligent and won't restart the service more than it needs to. I ran this against encoding01 and encoding02 and it consistently applied the manifest and restarted the workers with the --no-block option.

Also, I added an explicit daemon-reload since Puppet didn't seem to consistently do that when unit changes were made with file, but that's a fix for an edge case since that file will not change often. It's something that happened to me while testing that I figured I would fix.