Precompiled 2025-11-25-f765835

Use plain FORCE INDEX in PREV_REV_CHUNK_QUERIES (#43551)

GitOrigin-RevId: 1fb6b66e1d07b24a714a50a5d81e9feacc132a61

Precompiled 2025-11-25-4c81e5f

Only use CACHE_BREAKER_ARGS for componentQuery (#43522)

GitOrigin-RevId: f6125828933b2f643d470a0c161c757f34669750

Precompiled 2025-11-25-109abc7

dashboard: revert experimental pagination changes (#43502)

GitOrigin-RevId: a882bbacf247e4054daeb7483cc3287687f4cb2e

Precompiled 2025-11-25-049493b

load generator: fix scenarios (#43517)

GitOrigin-RevId: ebac0d547d08638b426ebc86fdd1068e24990813

Precompiled 2025-11-24-e50aa0a

codemod: fix build process to handle import alias (#43476)

GitOrigin-RevId: f27ab9257d98d43e8689a401b9aa3276fda8e46e

Precompiled 2025-11-22-73af2f4

Stop using a FollowerRetentionManager in database bootstrap (#43443)

Using FollowerRetentionManager needlessly makes database initialization (`DatabaseSnapshot::load`)
and retention workers slower because they repeatedly query persistence globals to get the
current snapshot timestamps, even though we hold the lease and are the only writer.

We can instead make everything use LeaderRetentionManager by splitting up its initialization into
two parts - one that reads the initial snapshot bounds & returns a usable RetentionValidator,
then a second part that actually launches the retention workers.

GitOrigin-RevId: cb8aef3f6349eafdd0efd73115b6bffcf242689e

Precompiled 2025-11-21-717fe22

Classify invalid component path errors. (#43409)

This classifies them as user errors - to propagate error back to client

GitOrigin-RevId: 79e7565ca24bc5e2eb85c55edd461d16e5e3f0a2

Precompiled 2025-11-21-4cadbb2

Up the retries for aws storage (#43432)

GitOrigin-RevId: a7b347f83c0d35916a05f159db96897c1307df48

Precompiled 2025-11-20-ef4ee69

fivetran_source: increment version for breaking changes (#43390)

GitOrigin-RevId: f7c8fa34ad63855cbbb05ff341b86eba220fa1a8

Precompiled 2025-11-20-90993c8

Ratelimit retention checkpointing every 5 min (#43375)

Update retention checkpointing rate limit to be every 5 min instead of every minute, and make the knob lowerable by setting it to a duration before the next  checkpoint instead of number of times per minute to checkpoint (we were already at the min of once per minute).

GitOrigin-RevId: be350f72962ab7982ac5aafb7b968fd11c5816bd