Move :katrix-store-sqlite:jvmTest to the quiet CI phase (D33 family: BUSY #257/#259, MigrationTest 60s runTest timeout #261) #102

New issue

Closed

opened 2026-08-21 12:25:11 +02:00 by thecrealm · 0 comments

thecrealm commented

2026-08-21 12:25:11 +02:00

Owner

Three phase-1 sightings on 2026-08-21 alone, all in katrix-store-sqlite while cargo/android/compose tasks ran beside it: ConcurrentWriterTest SQLITE_BUSY (build runs #257, #259 — stack on #74: BUSY at BEGIN via JdbcDriver.setAutoCommit), and MigrationTest v1_store_migrates_to_current_keeping_data_and_gaining_queue_and_cache hitting runTest's 60 s UncompletedCoroutinesError (#259... run #261, 2c2ea84). Two 30 s busy waits already exhaust a 1-minute runTest. D33's rule is serialize, don't bump waits: add :katrix-store-sqlite:jvmTest to the phase-2 list (D33 addendum 2). #74 keeps the driver-seam question (why BUSY escapes the busy handler at all).

Three phase-1 sightings on 2026-08-21 alone, all in katrix-store-sqlite while cargo/android/compose tasks ran beside it: ConcurrentWriterTest SQLITE_BUSY (build runs #257, #259 — stack on #74: BUSY at BEGIN via JdbcDriver.setAutoCommit), and MigrationTest v1_store_migrates_to_current_keeping_data_and_gaining_queue_and_cache hitting runTest's 60 s UncompletedCoroutinesError (#259... run #261, 2c2ea84). Two 30 s busy waits already exhaust a 1-minute runTest. D33's rule is serialize, don't bump waits: add :katrix-store-sqlite:jvmTest to the phase-2 list (D33 addendum 2). #74 keeps the driver-seam question (why BUSY escapes the busy handler at all).

thecrealm referenced this issue from a commit

2026-08-21 12:25:36 +02:00

CI: sqlite store tests run in the quiet phase (D33 addendum 2, closes #102)

thecrealm closed this issue

2026-08-21 12:25:36 +02:00

thecrealm referenced this issue

2026-09-11 21:19:28 +02:00

OfflineClientEndToEndTest.run_two_paints_from_the_store_before_any_network flakes under phase-1 load (CI #420, 20 s timeout) #174