Reorder install.sh to update docker images before shutting down

[giggsey]

Problem Statement

The install.sh script shuts down Sentry before running the 'Fetching and updating docker images' step.

On slow internet connections, this can cause quite a long downtime for Sentry, when the script is doing something unrelated.

Solution Brainstorm

Could the docker images be pulled before Sentry has been shutdown?

[aldy505]

There is a --minimal-downtime flag though

self-hosted/install/parse-cli.sh

Lines 11 to 16 in 0d73a94

	--minimize-downtime EXPERIMENTAL: try to keep accepting events for as long
	as possible while upgrading. This will disable cleanup
	on error, and might leave your installation in a
	partially upgraded state. This option might not reload
	all configuration, and is only meant for in-place
	upgrades.

But I think it would be great if this is implemented by default tho.

@aminvakil what do you think?

[aminvakil]

Could the docker images be pulled before Sentry has been shutdown?

Nice idea! I honestly I always execute docker compose pull --ignore-pull-failures manually, but your solution is much better, can you create a PR making this change?

There is a --minimal-downtime flag though

--minimize-downtime keeps relay and nginx running but it will stops other containers anyway. I think pulling docker images prior to stopping containers is a better default anyway.

[linear]

SELF-89

[mzglinski]

Hi @aldy505, I saw this issue and I decided that it might be a good place to mention that --minimize-downtime might not work that well any more after the intrudtionon of healthchecks. I've tried it recently and could not get the up (first one in the script) command to pass successfully. There was always some container that failed the startup procedure, do not know why, did not investigate further. Yet the default path works without problems, just thought it might be worth mentioning.