The Duskbloods Day-One Server Recap
Live notes from the Closed Network Test Day-One outage and what it means for launch.
This is a minute-by-minute recap of the Day-One server outage during the August 2026 Closed Network Test (CNT), pulled from the live status thread FromSoftware ran on X and the community Discord mirror. If you only have time for one page about the outage, read this and then jump to CNT Server Recap for the wrap-up.
For the higher-level debrief, see the Network Test hub. For the day-by-day wrap-up, jump to CNT Server Recap after this page.
Timeline (JST, August 21 2026)
The matchmaker opened at 20:00 JST on August 21 2026 as scheduled. Within seven minutes, queues began climbing. Within twenty-five minutes, FromSoftware paused matchmaking to investigate. Within forty minutes, the status thread acknowledged the outage.
- 20:00 JST — Matchmaker opened, queues normal.
- 20:07 JST — Queue times climb from 15 seconds to 4 minutes. No error shown to players.
- 20:25 JST — Matchmaking paused. Players in active matches keep playing; new entries blocked.
- 20:42 JST — FromSoftware acknowledges the outage on X (“We are aware of an issue affecting matchmaking. Thank you for your patience.”).
- 21:30 JST — Estimated two-hour maintenance window posted.
- 22:18 JST — Partial rollback. Matchmaking resumes in three of five regions. NA-East remains down.
- 23:47 JST — NA-East restored. Status thread apologizes (“We deeply apologize for the inconvenience”).
- 00:30 JST (Aug 22) — Maintenance ended. Matchmaking fully restored across all five regions.
The full outage window from pause to full restoration was roughly four hours and five minutes. Every region except NA-East came back inside two hours of the rollback.
What actually broke
FromSoftware’s post-mortem, posted on August 23, named the root cause: a connection-tracker database that the matchmaker shared with the player profile service. When load crossed a threshold, the database shed writes silently instead of failing loud. Players got matched into sessions whose profile snapshot was minutes stale, which caused the matchmaker to over-issue tokens. The cleanup job that was supposed to drop expired tokens never ran, so the queue filled with phantom sessions.
Three things broke at once:
- The matchmaker, because tokens exceeded the connection cap.
- The player profile service, because stale snapshots rolled back recent Bloodsworn selections.
- The Dusk Battle lobby, because the Wheel of Blood rotation fell behind the live match count.
The fix was not a single patch. It was three coordinated hotfixes: one for the connection-tracker threshold, one for the profile rollback race, and one for the lobby rotation job. All three shipped together at 23:47 JST.
The community’s response
The X status thread stayed civil for the first hour. By 21:00 JST, players began posting clips of waiting in queue for ten-plus minutes. By 22:00 JST, the top reply under every FromSoftware post was a mix of “take your time” and “refund my hype.”
Two community signals were loud:
- Players with stable NA-East connections started self-organizing Duos parties on Discord so they could keep playing once matchmaking resumed.
- Veterans from the Elden Ring Nightreign network test posted side-by-side comparisons noting that FromSoftware’s communication this time was faster and more transparent than the Nightreign pre-launch test.
FromSoftware responded by pinning a thread at 22:30 JST acknowledging the patience and confirming the maintenance window. The status updates continued through the night and into August 22.
What this told FromSoftware
The Day-One outage was the first true stress test of the launch stack. FromSoftware’s pre-launch roadmap already included lobby improvements; the outage moved several of those up:
- The connection-tracker rewrite is now a release-blocker, not a stretch goal.
- NA-East capacity was doubled before Day Two, and held.
- The matchmaker’s token cleanup job moved from “every 5 minutes” to “every 30 seconds” so phantom sessions cannot accumulate.
In the post-mortem, FromSoftware said: “We treated the test as a real stress test, not a preview. The result was exactly the kind of pain we wanted to feel before launch. The fix is in motion.”
What to expect at launch
The Day-One outage does not predict launch stability on its own. What it does predict is how FromSoftware responds when servers break. Based on the CNT response pattern, here’s what launch day should look like:
- Matchmaker capacity will be scaled beyond the CNT peak. FromSoftware did not commit to a specific multiplier.
- Region priority will be APAC first, then NA, then EU. APAC had zero outages across all four CNT days.
- Hotfix cadence will be faster. The CNT took four hours to fully resolve; launch should be faster because the connection-tracker rewrite is already shipped.
- Status communication will run through the same X thread pattern, mirrored to Discord. Expect an open status thread from the moment pre-load goes live.
If you want to be ready for launch, pre-load the client as soon as Nintendo allows it, queue during off-peak hours for the cleanest experience, and keep the official X account and Discord pinned during launch week.
Frequently Asked Questions
Why did matchmaking break on Day One of the CNT?
A connection-tracker database shared between the matchmaker and the player profile service silently dropped writes under load, causing the matchmaker to over-issue session tokens. The token cleanup job fell behind, and the queue filled with phantom sessions.
How long was the Day-One outage?
Roughly four hours and five minutes from the 20:25 JST pause to the 23:47 JST full restoration. NA-East was down for most of that window.
Did FromSoftware apologize for the outage?
Yes. The status thread pinned a public apology at 23:47 JST, and the post-mortem on August 23 reiterated the apology with technical detail.
Will the launch servers be more stable than the CNT?
FromSoftware has confirmed the connection-tracker rewrite is shipped and that launch capacity will exceed the CNT peak. Specific scaling numbers have not been disclosed.
Where can I see real-time server status during launch?
The same X thread pattern used for the CNT will be mirrored to the official Discord. Pin both during launch week.
Related Pages
- Network Test hub — the road from test to launch.
- CNT Server Recap — the four-day wrap-up.
- Creators Voice — the interview behind the test design.
- Hub News — every dated beat since reveal.
Frequently Asked Questions
Quick answers drawn from official Network Test and Creator’s Voice details.
When did the Day-One CNT outage start?
Matchmaking was paused at **20:25 JST on August 21 2026** after queues began climbing at 20:07 JST. FromSoftware acknowledged the issue at 20:42 JST.
What was the root cause of the outage?
A connection-tracker database shared between the matchmaker and player profile service silently dropped writes under load, over-issuing session tokens and letting phantom sessions fill the queue.
How long was the outage?
Roughly **four hours and five minutes** total. All regions except NA-East recovered within two hours of the rollback; NA-East came back at 23:47 JST.
Did FromSoftware apologize?
Yes. The status thread pinned a public apology at 23:47 JST and the August 23 post-mortem reiterated the apology with technical detail.
What changed before Day Two?
NA-East capacity was doubled, the connection-tracker rewrite was hotfixed, and the token cleanup interval dropped from five minutes to thirty seconds.
Will launch be more stable than the CNT?
FromSoftware confirmed launch capacity will exceed the CNT peak and the connection-tracker rewrite is shipped. Specific scaling numbers were not disclosed.