🔄 Sync with upstream changes by h0lybyte · Pull Request #17 · KBVE/realtime

h0lybyte · 2025-10-13T06:09:38Z

Upstream Sync

This PR contains the latest changes from the upstream repository.

Changes included:

Synced from upstream/main
Auto-generated by upstream sync workflow

Review checklist:

Review the changes for any breaking changes
Check for conflicts with local modifications
Verify tests pass (if applicable)

This PR was automatically created by the upstream sync workflow

…upabase#1548)

…abase#1549)

…ase#1551)

* fix: reduce max_frame_size to 5MB * fix: fullsweep_after=100 on gen rpc pub sub workers --------- Co-authored-by: Eduardo Gurgel Pinho <eduardo.gurgel@supabase.io>

* Also tweak buckets to account all the way to 3000KB * Start tagging the payload size metrics with message_type. message_type can be presence, broadcast or postgres_changes

It uses a fork of Phoenix for time being * fix: count presence_diff events on MessageDispatcher * fix: remove traces from console during development

Also cache mix _build and deps

* fix: use LiveView stream for status page * fix: need full node name on localhost for tests * fix: cleanup * fix: add tests * fix: bump version * fix: cleanup syntax * fix: format

--------- Co-authored-by: blacksmith-sh[bot] <157653362+blacksmith-sh[bot]@users.noreply.github.com> Co-authored-by: Eduardo Gurgel Pinho <eduardo.gurgel@supabase.io>

Selective broadcast for Postgres Changes. We use the fact that we know where each subscription subscribed from and do a direct broadcast for each node instead of spamming the whole cluster. If the subscription node is not readily available (for whatever reason) we fallback to sending to the whole cluster.

…ransport (supabase#1563) Also set fullsweep_after for RealtimeChannel processes

* Use GenRpc on SubscriptionsChecker * Use GenRpc on PostgresCdcRls

Tweaked a few tests and increase max cases to 4

Validates the size of the track message sent and fails if exceeds the value set by `tenant.max_payload_size_in_kb`

supabase/gen_rpc#5

…node (supabase#1583)

When regional_broadcasting is true GenRpcPubSub uses a single node in each region to broadcast to the rest of the region. Nodes communicating inside the same region have a much lower latency.

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

This way we can truly have Watchdog and ReplicationConnection linked

Also use Tasks to ensure that they drop the used memory after aggregating such metrics

Comma-separated list of `key=value` pairs appended as `extra_label` query parameters on each metrics push (e.g., `region=us-east-1,env=prod`). Useful for label injection supported by systems like VictoriaMetrics. If not set, no extra labels are added.

…1770) Include db_host_resolved to tenant info Also minor things like renaming the module to use RealtimeWeb and use external_id instead of project_ref

* ci: update supabase-js for e2e tests in workflow * Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

…supabase#1775)

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

* Application.fetch_env! does an ETS lock on every single read and we do it across several high volume areas of the service * Map.drop removed as Jason already tackles this and this will prevent map copy * Always use gen_rpc as the communication method between cluster

edgurgel and others added 30 commits October 1, 2025 11:53

fix: move message queue data to off-heap for gen_rpc pub sub workers (s…

a72a835

…upabase#1548)

fix: rate limit Connect.lookup_or_start_connection on error only (sup…

353c142

…abase#1549)

fix: increase connect error rate window to 30 seconds (supabase#1550)

748398c

fix: set a lower fullsweep_after flag for GenRpcPubSub workers (supab…

92e7b59

…ase#1551)

fix: hardcode presence limit (supabase#1552)

6248e2b

fix: further decrease limit on presence events (supabase#1553)

e84ac08

fix: bump up realtime (supabase#1554)

13052aa

fix: lower rate limit to 100 events per second (supabase#1556)

6e650f0

fix: move connect rate limit to socket (supabase#1555)

05ac93e

* fix: reduce max_frame_size to 5MB * fix: fullsweep_after=100 on gen rpc pub sub workers --------- Co-authored-by: Eduardo Gurgel Pinho <eduardo.gurgel@supabase.io>

fix: collect global metrics without tenant tagging (supabase#1557)

e9eaf9f

feat: presence payload size (supabase#1559)

16bd44d

* Also tweak buckets to account all the way to 3000KB * Start tagging the payload size metrics with message_type. message_type can be presence, broadcast or postgres_changes

fix: use GenRpc for Realtime.Latency pings (supabase#1560)

07de665

Fastlane for phoenix presence_diff (supabase#1558)

ecac071

It uses a fork of Phoenix for time being * fix: count presence_diff events on MessageDispatcher * fix: remove traces from console during development

fix: limit db events (supabase#1562)

058be58

chore: split tests and lint workflows (supabase#1564)

d5658ad

Also cache mix _build and deps

fix: use LiveView stream for status page (supabase#1565)

8b621bd

* fix: use LiveView stream for status page * fix: need full node name on localhost for tests * fix: cleanup * fix: add tests * fix: bump version * fix: cleanup syntax * fix: format

fix: refine join payload checking (supabase#1567)

aeafab6

fix: shard user scopes in syn (supabase#1566)

cbcbbfd

chore: migrate workflows to Blacksmith runners (supabase#1569)

05ec589

--------- Co-authored-by: blacksmith-sh[bot] <157653362+blacksmith-sh[bot]@users.noreply.github.com> Co-authored-by: Eduardo Gurgel Pinho <eduardo.gurgel@supabase.io>

fix: set max_heap_size to RealtimeChannel the same as the websocket t…

c11f0a4

…ransport (supabase#1563) Also set fullsweep_after for RealtimeChannel processes

fix: use gen_rpc on Postgres Changes (supabase#1577)

fc46833

* Use GenRpc on SubscriptionsChecker * Use GenRpc on PostgresCdcRls

chore: more async tests (supabase#1578)

28d8609

Tweaked a few tests and increase max cases to 4

fix: validate size of track message (supabase#1574)

c9ced7c

Validates the size of the track message sent and fails if exceeds the value set by `tenant.max_payload_size_in_kb`

fix: set sane defaults for presence event limit (supabase#1544)

617afae

feat: bump up to 1.18 and otp28 (supabase#1582)

a125ba1

fix: update gen_rpc with fullsweep_after=20 (supabase#1581)

9541666

supabase/gen_rpc#5

fix: on ReplicationPoller send only subscription IDs relevant to the …

0ffb34b

…node (supabase#1583)

feat: regional broadcasting (supabase#1580)

f44287e

When regional_broadcasting is true GenRpcPubSub uses a single node in each region to broadcast to the rest of the region. Nodes communicating inside the same region have a much lower latency.

chore: remove demo (old multiplayer.dev) (supabase#1586)

9adde30

supabase-libs-pr-manager bot and others added 30 commits March 24, 2026 08:40

chore: update @supabase/supabase-js to v2.100.0 (supabase#1768)

23593c3

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

fix: stop trapping exists on ReplicationConnection (supabase#1767)

a81ae8a

This way we can truly have Watchdog and ReplicationConnection linked

chore(release): 2.78.17 [skip ci]

3e41a38

fix: always send standby status to Postgres (supabase#1769)

ee5b5b9

chore(release): 2.78.18 [skip ci]

bd7cbf6

fix: change MetricsPusher to send tenant metrics (supabase#1759)

5d80128

Also use Tasks to ensure that they drop the used memory after aggregating such metrics

chore(release): 2.78.19 [skip ci]

6e36795

chore(release): 2.79.0 [skip ci]

b78bca3

feat: Include resolved db address to tenant info dashboard (supabase#…

4c98194

…1770) Include db_host_resolved to tenant info Also minor things like renaming the module to use RealtimeWeb and use external_id instead of project_ref

chore(release): 2.80.0 [skip ci]

3fa05b6

chore: update @supabase/supabase-js to v2.100.1 (supabase#1772)

c68e4c4

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

ci: include changelogs in js update pr (supabase#1773)

681efc6

fix: detect local ipv6 host and stop (supabase#1774)

bc21971

chore(release): 2.80.1 [skip ci]

16b8dd6

fix: use ECS_CONTAINER_METADATA_URI_V4 to grab public ipv6 on fargate (…

866abf9

…supabase#1775)

chore(release): 2.80.2 [skip ci]

4f96ab8

chore: update @supabase/supabase-js to v2.101.0 (supabase#1777)

d78d5b7

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

fix: add random jitter to replication poller (supabase#1780)

c7f0c0d

chore(release): 2.80.3 [skip ci]

da09177

chore: update @supabase/supabase-js to v2.101.1 (supabase#1779)

fff52d1

Co-authored-by: supabase-workflow-trigger[bot] <266661614+supabase-workflow-trigger[bot]@users.noreply.github.com>

fix: reduce map heap size (supabase#1781)

e70077f

chore(release): 2.80.4 [skip ci]

ed03011

fix: remove billing metrics join logs (supabase#1783)

90cc03a

chore(release): 2.80.5 [skip ci]

9f8258f

chore(release): 2.80.6 [skip ci]

94f5139

fix: reduce channel rate limit log cadence (supabase#1784)

4bc7a38

chore(release): 2.80.7 [skip ci]

22c3151

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

🔄 Sync with upstream changes#17

🔄 Sync with upstream changes#17
h0lybyte wants to merge 271 commits intomainfrom
upstream-main

h0lybyte commented Oct 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

10 participants

Conversation

h0lybyte commented Oct 13, 2025

Upstream Sync

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

10 participants