[compute_ctl] Create check availability data during full configuration (#5084)

I've moved it to the API handler in the 589cf1ed2 to do not delay
compute start. Yet, we now skip full configuration and catalog updates
in the most hot path -- waking up suspended compute, and only do it at:

- first start
- start with applying new configuration
- start for availability check

so it doesn't really matter anymore.

The problem with creating the table and test record in the API handler
is that someone can fill up timeline till the logical limit. Then it's
suspended and availability check is scheduled, so it fails.

If table + test row are created at the very beginning, we reserve a 8 KB
page for future checks, which theoretically will last almost forever.
For example, my ~1y old branch still has 8 KB sized test table:
```sql
cloud_admin@postgres=# select pg_relation_size('health_check');
 pg_relation_size
------------------
             8192
(1 row)
```

---------

Co-authored-by: Anastasia Lubennikova <anastasia@neon.tech>
This commit is contained in:
Alexey Kondratov
2023-08-30 17:44:28 +02:00
committed by GitHub
parent 93dcdb293a
commit f2c21447ce
3 changed files with 41 additions and 10 deletions

View File

@@ -138,7 +138,13 @@ impl ComputeControlPlane {
mode,
tenant_id,
pg_version,
skip_pg_catalog_updates: false,
// We don't setup roles and databases in the spec locally, so we don't need to
// do catalog updates. Catalog updates also include check availability
// data creation. Yet, we have tests that check that size and db dump
// before and after start are the same. So, skip catalog updates,
// with this we basically test a case of waking up an idle compute, where
// we also skip catalog updates in the cloud.
skip_pg_catalog_updates: true,
});
ep.create_endpoint_dir()?;
@@ -152,7 +158,7 @@ impl ComputeControlPlane {
http_port,
pg_port,
pg_version,
skip_pg_catalog_updates: false,
skip_pg_catalog_updates: true,
})?,
)?;
std::fs::write(