Elasticsearch mapping schema drift — immutable field type mismatch

Now let me invoke the RCA skill.

Let me search Datadog logs and identify the repository in parallel.

Excellent. The Datadog search revealed this is an Elasticsearch mapping migration issue triggered by

Now I need to find the Record model's Elasticsearch mapping definition to see how `priority` is defi

Excellent findings. Now let me read the actual mapping file for Record to see the `priority` field d

Now I need to understand why the existing ES index has `priority` as `long` when the code defines it

Now let me find out how `update_mapping!` gets called on deployment. Let me check the deploy process

So `update_mapping!` is only explicitly called in migrations and `partial_reindex!`. Let me check ho

Now let me look for how `update_mapping!` gets called on deploy — checking for EB platform hooks or

The `update_mapping!` is not called on boot/deploy — it's only triggered by migrations and `partial_

Let me search for how `put_mapping` or mapping updates might be triggered. The error logs say the fu

Now let me check for the deploy-time mapping sync. Let me search for an Elasticsearch initializer or

No mapping sync on boot. Now let me search for the deploy script or hook that triggers `update_mappi

Now I see — there's an `ElasticsearchMigrationRunner`. Let me find the ES migration files and the mi

Now let me look for the deploy hooks that might call `es:migrate` or `update_mapping!` on deploy.

Found it. Let me read the deploy hook.

Wait — `es:create_index` only creates indices that don't exist. And `es:migrate` runs ES migration f

Now let me check the `tesla:init` rake task and look for anything that calls `update_mapping!` on al

The deploy flow is: `es:create_index` (only creates missing) -> `es:migrate` (runs ES migration file

No mapping sync on Sidekiq startup. The errors all hit simultaneously with a deploy. Let me re-exami

That's an old migration from 2023. Let me look for an `ElasticsearchMigration::Base` class and any E

So there are no ES migration files, no deploy-time `update_mapping!` call. The `update_mapping!` is

Let me check specifically for a Sidekiq startup hook or a puma `on_worker_boot` hook.

Let me focus on the actual evidence from Datadog logs. The error function is `update_mapping!` and i

I now have sufficient evidence to write the RCA. The key finding is:

Now let me invoke the RCA skill.

The error is: `mapper [priority] cannot be changed from type [long] to [integer]` — an Elasticsearch

Excellent findings from both agents. Let me now read the key source files to get exact code snippets

Now let me find the cron searchable code and check for the recent commit that may have changed the m

Now let me check the git log for the specific deployment that introduced this issue — the `20260503T

No recent commits to these files — meaning the mapping definitions (`type: 'integer'`) have been in

This is the key finding. The cron schedule at `config/schedule.rb:148` runs `Cupix::Cron::Searchable

Important detail: `update_mapping` (without `!`) at line 295 calls `update_mapping!` and **rescues**

The directory and file exist. Let me read the existing file first since I need to overwrite it.

There's already an existing RCA report, but it has a significant gap — the Datadog log search found

Now let me update the cluster frontmatter. Per instructions, I should NOT modify the frontmatter — t

RCA report written to `./content/docs/incidents/e5a7c69a-d146-4560-b89f-7ddf7efc46dd/rca.mdx`.

RCA complete. The report has been written to `./content/docs/incidents/e5a7c69a-d146-4560-b89f-7ddf7