• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

supabase / realtime / 43131474481
93%

Build:
DEFAULT BRANCH: main
Ran 12 Nov 2026 01:06AM UTC
Jobs 1
Files 142
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

12 Nov 2026 01:00AM UTC coverage: 93.133% (-0.06%) from 93.191%
43131474481

push

github

web-flow
chore(rollcall): rollcall distributed tests (#2547)

* chore(rollcall): setup env vars

* chore(rollcall): distributed tests for rollcall

* chore(rollcall): pin the presence flags in the flag cache in tests

The flag cache refills an expired entry in a process of its own. In an async
test that process cannot read the sandbox, so the refill fails and the flag
reads as off. A test in the Rollcall mode then falls back to Phoenix in the
middle of the test.

Insert the two presence flags without Api.upsert_feature_flag/1, because it
updates the cache with a cast that lands later and gives the entry an expiry
again. Put them in the cache with no expiry.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* chore(rollcall): set the Rollcall timing for tests in test.exs

Remove the ROLLCALL_* env vars and their conditions from runtime.exs, and the
rows of ENVS.md. Rollcall is not used in production yet, and the only need is
a shorter timing in the tests. Set the options in config/test.exs. The
application reads them from the app env, and the peer nodes of the tests copy
the app env. Without the key, the library defaults apply.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* test(rollcall): return only the socket from connect in the distributed tests

A test that does not need the presence_state no longer matches a 2-tuple. The
two tests that read the presence_state of a client that joins later call
ClusterPresence.connect_client/4 directly.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* test(rollcall): extract the listener and tracker setup of the distributed tests

Five tests started with the same lines: a listener on node A, a tracker on the
test node, a track and the wait for the join. A helper now does this, so each
test shows only what makes it different.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

4801 of 5155 relevant lines covered (93.13%)

4708.04 hits per line

Coverage Regressions

Lines Coverage ∆ File
2
90.24
-4.88% lib/realtime/monitoring/peep/partitioned.ex
1
93.33
-6.67% lib/realtime/presence/rollcall_transport.ex
1
91.05
-0.53% lib/realtime/tenants/connect.ex
1
86.91
-0.52% lib/realtime/tenants/replication_connection.ex
Jobs
ID Job ID Ran Files Coverage
1 43131474481.1 12 Nov 2026 01:06AM UTC 568
62.38
GitHub Action Run
Source Files on build 43131474481
  • Tree
  • List 142
  • Changed 118
  • Source Changed 0
  • Coverage Changed 118
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #43131474481
  • adbadd5d on github
  • Prev Build on main (#43114163261)
  • Next Build on main (#43145039069)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc