Migrate to humantime (phase 1)

[squadgazzz]

Description

Closes phase 1 from the #2085. Since DEX solvers are about to migrate to a separate repo, it's a good time to change the config format. It will be much more difficult later.

Changes

Support both plain and humantime formats to migrate our Pulumi config gradually.

How to test

e2e + staging

[fleupold]

Looks good, just a small confusion around the continuation of "from_days/from_seconds"

[fleupold]

Can we not remove this method in favor of humantime::parse_duration altogether? I think it's weird to have a method duration_from_day that you can call with "10m"

Or is this phase2 after the pulumi config update?

[squadgazzz]

Or is this phase2 after the pulumi config update?

Yes, this method will be either removed or renamed since s.parse::<f64>().map(|days| Duration::from_secs_f64(days * 86_400.0)) part will be gone for sure.

[MartinquaXD]

I think I would not switch over in multiple phases.
The infra PR that changes from the old config to the new one can be prepared before hand and verified locally. And when this PR gets merged it can quickly be applied.
When we apply the changes together in prod next week we'll have no trouble.
So I think the only upside to splitting this change into multiple phases is avoiding a tiny downtime on staging which would be totally okay. It's a testing environment which is allowed to break from time to time.

[squadgazzz]

When we apply the changes together in prod next week we'll have no trouble.

I was worried about that exactly. Why there won't be downtime as we would have in staging? infra and services are deployed at the same time in prod?

The infra PR that changes from the old config to the new one can be prepared before hand and verified locally.

Not sure how to test it locally before merging this PR. Applying the new config will fail until this PR is not on staging.

[MartinquaXD]

Why there won't be downtime as we would have in staging? infra and services are deployed at the same time in prod?

Exactly.

Not sure how to test it locally before merging this PR.

You could probably log the content of the files pulumi would create if you were to apply the change and try to run your local system with those to see if things break. But that might be a bit involved if you want to make sure everything works.

Alternatively it should be possible to build a docker container from the PR branch and use that when preparing your infra change. Then you can play around with both changes together on sepolia for example until they work together.

IIRC in order to make CI build a container from your branch it should have a tag starting with v (see CI).

[squadgazzz]

#2194