Conversation
Document Engine 1.19.0 makes its remote-URL downloader configurable, so the chart needs the matching values. Without them a customer following the reference architecture cannot reach the new settings at all. Adds `config.urlFetchPoolSize`, `config.urlFetchQueueTimeoutSeconds` and `config.urlFetchQueueMaxSize`, mapped to `DOWNLOADER_WORKER_POOL_SIZE`, `DOWNLOADER_BROKER_CLIENT_QUEUE_TIMEOUT` and `DOWNLOADER_BROKER_CLIENT_QUEUE_MAX_SIZE`. Defaults match what Document Engine used while these were hardcoded, so an existing deployment is unaffected. Companion to https://linear.app/nutrient/issue/SER-2815
This is clearly coming from an agent, which is fine, but it also shows that you don't have all the tools required installed. Install them and generate the schema, do not hand edit. Any subsequent generation can produce more changes, make following PRs unnecessary big. We should also put this PR on hold, add a TODO item here, and merge when DE 1.19 is out (unless we want to have this option available in |
The chart has never gated an entry on a Document Engine version. This PR waits for the 1.19.0 release commit, which bumps appVersion first, so the requirement holds by ordering and the entry can read like every other feature entry.
|
Holy! Yea, I didn't know about this at all. First time to touch such changes. Tried to weigh it right with the agent but apparently failied 😅 Many thanks for the correction, @zrzka All three done. Schema is generated, PR back to draft, TODO is on #58563 Waiting for 1.19 rather than nightly. I set both variables on released 1.18.1 and nothing changed, still 16 and shedding at 5s, so merging early would just hand people a dead knob. Is that the right way? Also put up #168 for your review. It's for the regeneration command part. |
Companion to https://linear.app/nutrient/issue/SER-2815 and the Document Engine PR.
Summary
Document Engine 1.19.0 makes its remote-URL downloader configurable. The reference architecture points customers at this chart, so without these values the feature is unreachable for anyone deploying that way.
Defaults match what Document Engine used while these were hardcoded, so an existing deployment is unaffected.
On hold until Document Engine 1.19.0 is out. Tracked as a release TODO in the monorepo.
Two things for whoever merges this. The 1.19.0 release lands as its own commit that bumps appVersion and takes the next chart minor, which by the pattern of 1.16.0 to 8.4.0, 1.17.0 to 8.5.0 and 1.18.0 to 8.6.0 will be 8.7.0. So this PR needs renumbering to 8.8.0 once that lands. And the changelog entry deliberately carries no "requires 1.19.0" note, because merging after the appVersion bump makes it true by ordering and the chart has never version-gated an entry.
Fix
values.yaml-config.urlFetchPoolSize,config.urlFetchQueueTimeoutSeconds,config.urlFetchQueueMaxSize, next to the existingurlFetchTimeoutSecondstemplates/configmap.yaml- maps them, seconds to milliseconds like its neighboursChart.yaml- 8.6.3 to 8.7.0, backward compatible additionCHANGELOG.md,README.md(helm-docs),values.schema.json(helm schema)QA
helm templatewith--set config.urlFetchPoolSize=32rendersDOWNLOADER_WORKER_POOL_SIZE: "32", and 5 seconds rendersDOWNLOADER_BROKER_CLIENT_QUEUE_TIMEOUT: "5000"--set config.urlFetchPoolSize=0is rejected by the schema, matching the validation in Document Enginehelm lintpasses