[CPDLP-3521] Split contract migration into two migrators

[leandroalemao]

Context

Ticket: https://dfedigital.atlassian.net/browse/CPDLP-3521

We're seeing ERROR: cached plan must not change result type when migrating some contracts over.

Changes proposed in this pull request

It's a concurrency issue, where a contract template is being created then a linked contract is being initialised multiple times as we have multiple workers trying to create the same contract. We're splitting contract migrator into two new migrators (contract template and contract). Also limiting the workers # for the contract migrator.

[github-actions]

Review app deployed to https://npq-registration-review-1819-web.test.teacherservices.cloud/

[cwrw]

I think we shouldn't be picking up the same records through multiple workers because of this: https://github.com/DFE-Digital/npq-registration/blob/main/app/services/migration/migrators/base.rb#L60
However the interesting thing is here: https://github.com/DFE-Digital/npq-registration/blob/main/app/services/migration/migrators/base.rb#L54 this might be adding the extra records you mentioned as it's doing a ceiling on both, maybe we can discuss this. Let's chat tomorrow with @ethax-ross. Nice catch, thanks @leandroalemao.

Nvm I just reran and it raised the same failure with 1 worker, let's discuss tomorrow

[ethax-ross]

Could we just wrap the contract creation in a transaction to avoid the race conditions? alternatively I think instead of overriding the constant you could just do something like this I think?:

def number_of_workers
  1
end

def records_per_worker
  record_count
end

[leandroalemao]

Hi @ethax-ross tried that yesterday.. but it breaks some shared specs..

[sonarcloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[leandroalemao]

Closing this for now. More info on https://dfedigital.atlassian.net/browse/CPDLP-3521.