docs(actors): sync from rivet-dev/rivet - #118
Conversation
| skill: true | ||
| --- | ||
|
|
||
| Expose [Prometheus metrics](/actors/docs/environment-variables#metrics) on your worker to collect these metrics. |
There was a problem hiding this comment.
🟠 Medium · Link metric setup to an existing section
This target has no #metrics heading, so the guide's prerequisite link lands on the environment-variable page without explaining how to expose or scrape metrics. Readers cannot complete the setup required for every PromQL example below. Link to the existing worker Prometheus setup page used by sqlite-profiling.mdx, or add the missing Metrics section before pointing here.
|
|
||
| #### Page cache capacity | ||
|
|
||
| `RIVETKIT_SQLITE_OPT_VFS_PAGE_CACHE_CAPACITY_PAGES` (default: `50000` pages) |
There was a problem hiding this comment.
🟠 Medium · Document the worker-wide memory multiplier
This environment variable is read once per process, but the configured capacity is applied to each Actor's page cache. Increasing it on a worker hosting many SQLite Actors can therefore multiply memory use across all active caches, even though the surrounding experiment is framed around one Actor/query. State that tuning requires an isolated representative worker and that the production memory budget must account for concurrent Actor caches; also include the implementation's 500,000-page cap so operators know the effective range.
Automated docs sync from
rivet-dev/rivet@a1c18da.Do not edit
vendor/actors/here. Edit the docs inrivet-dev/rivetand this PR updates itself.