Conversation
armru
force-pushed
the
dev/135
branch
3 times, most recently
from
September 2, 2026 08:08
af2a787 to
82b80a3
Compare
gabriele-wolfox
force-pushed
the
dev/135
branch
3 times, most recently
from
September 3, 2026 13:38
e9eac37 to
6d4baea
Compare
The work queue can now live either on its own volume or inside the tier1 data volume, so the server has to relocate it when the Server resource changes which one it is. QUEUE_MIGRATION_SOURCE names the previous location: its content is copied into QUEUE_DIRECTORY before the queue is opened, then removed. The copy is staged next to the destination and fsynced before the rename, so an interrupted migration never leaves a partial queue behind. A non-empty destination always wins, as it holds the WAL files still pending transfer to tier2. Signed-off-by: Armando Ruocco <armando.ruocco@enterprisedb.com>
The queue only needs durable storage, not a volume of its own, so the `queue` section is no longer required alongside tier1: when omitted, the queue lives in the `queue` directory of the tier1 data volume. A dedicated volume is still recommended in production, so that a full data volume cannot also stall the queue that drives retention. Adding or removing the section moves the queue during the resulting rolling restart. On removal the StatefulSet stops templating the queue PVC, which is then kept mounted by claim name until it is deleted by hand. Signed-off-by: Armando Ruocco <armando.ruocco@enterprisedb.com>
Adding a dedicated queue volume to an existing Server crash looped it: copyQueue replaced destination with a rename, assuming it shared a filesystem with the staging copy next to it, but destination is itself a mount point in that case, so removing or renaming into it fails. Stage the copy inside destination instead, then adopt each entry with its own rename, which stays on destination's filesystem either way. TestMigrateQueueDirectory/moves_the_queue_into_a_freshly_mounted_destination reproduces this with a real bind mount and now passes; it failed with "device or resource busy" before this change. Assisted-by: Claude Signed-off-by: Gabriele Quaresima <gabriele.quaresima@enterprisedb.com>
gabriele-wolfox
force-pushed
the
dev/135
branch
from
September 3, 2026 14:57
6d4baea to
6c7258e
Compare
gabriele-wolfox
approved these changes
Sep 3, 2026
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
2 participants
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.
A Server had to declare a queue PVC whenever tier1 was configured. The queue
only needs durable storage, not a volume of its own, so the
queuesection isnow optional: when omitted, the queue lives in the
queuedirectory of thetier1 data volume. A dedicated volume is still recommended in production, so
that a full data volume cannot also stall the queue that drives retention.
Adding or removing the section moves the queue during the resulting rolling
restart: the server copies it from the previous location, named by
QUEUE_MIGRATION_SOURCE, before opening it. On removal the StatefulSet stops
templating the queue PVC, which is then kept mounted by claim name until it is
deleted by hand.
Closes #135