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

tari-project / tari / 35102703514
74%

Build:
DEFAULT BRANCH: development
Ran 16 Sep 2026 02:23PM UTC
Jobs 1
Files 779
Run time 2min
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

16 Sep 2026 01:33PM UTC coverage: 62.884% (-0.008%) from 62.892%
35102703514

push

github

web-flow
test: stop comms tests dialling links something else already dials (#8008)

Both tests dial a peer that is already being dialled, and the redundant
connection is tie-broken against the live one. The dialer de-duplicates
only dials that are still in flight, and `handle_dial_peer` decides from
the connectivity pool, so a dial issued after the first has completed
but before `PeerConnected` reached the connectivity actor is issued
again.

Outbound messaging does requeue a failed write, but `MAX_SEND_RETRIES`
is 1: after two failed `try_establish` attempts `run_inner` calls
`fail_all_pending_messages`, which does not requeue. The messages are
gone and the test waits for one that will never arrive.

test_dht_propagate_dedup dialled each link itself, on top of every
node's DHT connectivity already dialling its seeded peers at start-up -
that goes via ProactiveDialer, which bypasses handle_dial_peer and does
not even log. Every link was dialled twice from the same side, which is
why the tie-breaks were all (Existing = Outbound, New = Outbound). Node
A is now seeded with both B and C, and ensure_connected dials only a
link that has not come up within DIAL_FALLBACK_GRACE. The fallback is
not optional: the DHT tries once at start-up and then not again until
update_interval (2 minutes), so one attempt lost under load hangs the
test. That failure mode only appeared under --release, after 450 clean
debug runs.

peer_to_peer_messaging_simultaneous let node 2 dial back because its
pool had not registered the inbound connection yet. Both pools are now
confirmed before either node sends.

Each node is waited on only for the peers it has to send to, which is
the pool lookup propagation actually performs - a receiver does not need
the connection in its own pool to be delivered to, and requiring that
only added a way for the barrier to time out on something unused.

Measured under 3x CPU oversubscription, the dedup test goes from 8-10
dials and 1-3 tie-breaks per run to z... (continued)

2 of 2 new or added lines in 1 file covered. (100.0%)

30 existing lines in 9 files now uncovered.

86491 of 137540 relevant lines covered (62.88%)

193159.97 hits per line

Coverage Regressions

Lines Coverage ∆ File
7
73.03
-1.97% comms/core/src/protocol/messaging/protocol.rs
5
87.2
-0.91% comms/core/src/connection_manager/peer_connection.rs
4
82.14
-0.79% comms/core/src/noise/socket.rs
4
89.57
-2.45% comms/core/src/protocol/messaging/outbound.rs
3
72.73
-0.08% comms/core/src/connectivity/manager.rs
3
92.18
-0.63% comms/dht/src/outbound/broadcast.rs
2
78.13
-3.13% comms/core/src/pipeline/outbound.rs
1
69.81
-0.24% base_layer/core/src/mempool/sync_protocol/mod.rs
1
83.33
-0.93% comms/core/src/protocol/messaging/inbound.rs
Jobs
ID Job ID Ran Files Coverage
1 35102703514.1 16 Sep 2026 02:23PM UTC 779
62.88
GitHub Action Run
Source Files on build 35102703514
  • Tree
  • List 779
  • Changed 17
  • Source Changed 1
  • Coverage Changed 17
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #35102703514
  • d7f3a8a9 on github
  • Prev Build on development (#35084441999)
  • Next Build on development (#35192540237)
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