Repository navigation
Run only the network-dependent Workers scenarios on the edge, at one date - #198
Merged
Merged
Conversation
…t one date The edge job ran most of the local workerd suite a second time against free-tier Neon and Supabase, at two compatibility dates, with timeouts sized for localhost. Nearly every edge failure that was not a driver change came from those repeated scenarios (Neon CONNECT_TIMEOUT while its compute woke, copy-out stalls, the timeouts in the 2026-10-23 network incident), and each was answered with another runner workaround. The edge now runs only what local workerd cannot show: diagnostics_channel in the production runtime, verified TLS and SNI against Neon, the private CA case, the poolers, the abort-* transactions, Hyperdrive and a client left open. arrays, timers, end-*, copy-out and subscribe run locally only. It runs at 2026-10-12 only. What the older dates change is close-time rejection timing, which the edge runner always tolerated and local workerd checks at every date, with and without the flag. Also removes the Neon wake request and the fetch-failure retry (reverts 7f0a439): with arrays-* off the edge, the first Neon connection has a 10 s or the default timeout, and a dropped runner fetch, once in about 400 jobs, is cheaper to re-run than to code around. A before/after run (4 jobs each at 294aa4c and 7f0a439) stalled copy-out once on each side, so that commit did not cause the stalls. Drops the unused known-failures settings in both runners, the hyperdrive_rounds input, and the date suffix on Preview names.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The edge job ran most of the local workerd suite a second time, against free-tier Neon and Supabase, at two compatibility dates, with timeouts sized for localhost. Almost every edge failure that wasn't a driver change came from those repeated scenarios, and each one got another workaround in the runner.
diagnostics-channel,ssl-verify,tls-private-ca,pooled,no-end, the sixabort-*and the threehyperdrive*.arrays-*,timers,end-*,copy-outandsubscriberun in local workerd only, at all four date/flag combinations.hyperdrive_roundsinput and the date suffix on Preview names.A before/after check (4 edge jobs each at 294aa4c and 7f0a439) stalled
copy-outonce on each side, so the stalls on main today are environmental, not from 7f0a439.The old
*-20250101and*-20261012Previews stay on the account until deleted by hand; nothing runs them.