Service notices
Planned maintenance, restarts, and outages for the shared relay, the archive, and this website. Newest first. This page is served from disk, so it keeps answering while the rest of the site is down.
· Planned maintenance
The archive and relay will be unavailable for up to three minutes while new production health metrics are activated. Connected worlds reconnect automatically. Participants do not need to act.
Update 2026-08-24 19:40 UTC. Service returned at 19:35:06 UTC. The participant outage ran from 19:33:55 to 19:35:06 UTC and lasted 71 seconds. The archive subscribed before participant traffic returned, and the production health metrics are active.
· Planned maintenance
The relay and live map will pause for up to five minutes while the archive restarts to deploy explicit population-gate limits and refusal/reroute animation. Participant worlds do not need to act; their sidecars reconnect automatically.
Update 2026-08-24 13:04 UTC. Service returned at 13:03:25 UTC. The participant outage lasted 88 seconds; the new live-map admission display and explicit refusal/reroute evidence are active.
· Planned maintenance
The archive and relay will be unavailable for up to three minutes. We will update chart inspection and the system map. Connected worlds reconnect automatically. Participants do not need to act.
Maintenance began at . Service returned at . The outage lasted 74 seconds.
· Planned maintenance
The archive and relay will be unavailable for up to three minutes while the production telemetry dashboard is activated. Connected worlds reconnect automatically. Participants do not need to act.
Update 2026-08-24 02:58 UTC. Service returned at 02:57:05 UTC. The participant outage ran from 02:55:52 to 02:57:05 UTC and lasted 73 seconds. The archive subscribed before participant traffic returned, and the production telemetry dashboard is active.
· Planned maintenance
The shared relay, archive, live map, and website will pause once shortly. The expected outage is about two minutes and no more than five while the archive starts a correction that shows a creature entering a world only after that world acknowledges delivery.
Your own world keeps running. Crossings pause while the relay is held down, the permanent record stays complete, and every world reconnects to its existing slot automatically. You do not need to do anything.
A follow-up line below will give the actual return time.
Update 2026-08-24 02:12 UTC. Service returned at 02:10:41 UTC. The participant outage lasted 104 seconds; the archive subscribed before any placement claim, the permanent record stayed complete, and the live map now animates only receiver-confirmed deliveries.
· Planned maintenance
The shared relay and website pause once between 21:28 and 21:35 UTC on 23 August 2026. The expected outage is about two minutes while the archive starts the release that explains population admission on the live map.
Your own world keeps running. Crossings pause while the relay is held down, the permanent record stays complete, and every world reconnects to its existing slot automatically. You do not need to do anything.
A follow-up line below will give the actual return time.
Update 2026-08-23 21:31 UTC. Service returned at 21:29:40 UTC. The participant outage lasted 78 seconds; the archive subscribed before any placement claim, and the permanent record stayed complete.
· Service terms
This notice is late. Release 0.3.0 changed this behavior in the participant sidecar. We did not post this notice with release 0.3.0.
Previous behavior. A creature left for an adjacent world. If no answer arrived, the sender sent the creature again. If no answer arrived before the old deadline, the sender returned the creature home automatically.
Behavior in release 0.3.0 or later. The sender sends the creature once. If no answer arrives before the configured deadline, the sender records the creature as lost. The sender does not send the creature again. It does not return the creature home.
Reason. If the neighbor receives the creature but sends no answer, the old retry can create two copies. A duplicated creature breaks the shared world. The map cannot repair it. The current rule accepts one lost creature to prevent an automatic duplicate.
What you will see. The live map shows a lost count for each world. Your local status shows unanswered crossings and recorded losses. An increase in the lost count can identify an unreliable connection to a neighbor. This count does not mean that your world duplicated a creature.
Participant action. Use the latest release for a new installation or an upgrade.
You can update an existing world later. If your sidecar is older than release 0.3.0, it uses the previous behavior until you update. World data and server settings do not require a manual change.
· Planned maintenance
The public archive, map, and shared relay can be unavailable for up to ten minutes while the archive rebuilds its Species genealogy. Connected sidecars will retry automatically. Participants do not need to act.
Update 2026-08-23 18:22 UTC. Service returned at 18:19:33 UTC. The outage lasted 10 minutes 24 seconds, which was 24 seconds longer than announced. The archive rebuilt the complete record, and the genealogy integrity checks passed.
· Planned maintenance
Website and map connections can be unavailable for up to eight minutes while we activate the latest installer guidance.
No participant action is required. Sidecars reconnect automatically. The relay stays held during the archive restart, so the permanent record remains complete.
Update 2026-08-22 03:25 UTC. Service returned at 03:22:28 UTC. The outage lasted 93 seconds. All checks passed.
· Planned maintenance
The website and public map can pause for up to 10 minutes. Worlds need no action.
This work rebuilds the recorded ancestry index from the complete archive.
Update 2026-08-22 00:22 UTC. Service returned at 00:16 UTC. The pause lasted 7 minutes 4 seconds. The ancestry index was rebuilt from the complete archive and now tracks every migrated species. The one living species without recorded ancestry is excluded seed stock that has never crossed between worlds.
· Planned maintenance
The shared relay, archive, live map, and main website will pause shortly. The expected outage is about two minutes.
Your world will keep running on your computer. You do not need to do anything. Its sidecar will reconnect automatically after service returns.
The relay will stay stopped while the archive reloads. This pause keeps the permanent crossing record complete.
The announcements page will stay available during the restart.
Update 2026-08-22 00:02 UTC. Service returned at 00:00:47 UTC. The interruption lasted 98 seconds. The archive subscribed before participant traffic returned, and the permanent record stayed complete. Sixteen worlds were live after the reconnection window. One previously dark slot remained dark.
· Planned maintenance
The website and live map will pause for about two minutes, and no more than three. This maintenance improves the Species view and applies a system update. Connected worlds retry automatically. No participant action is necessary.
Update 2026-08-21 20:57 UTC. The maintenance start moved to 21:10 UTC. The service remains available.
Update 2026-08-21 21:14 UTC. Maintenance is complete. The website and live map returned at 21:12:24 UTC. The pause lasted 2 minutes 20 seconds. The archive subscribed before the first world placement, so the crossing record is complete.
· Planned maintenance
The website, the live map and the broadcast page are unavailable for a short window in the next few minutes while the archive restarts. We expect about a minute, and no more than two. Worlds keep running and saving on their own machines throughout; nothing is lost, and no participant action is needed. Crossings pause for the length of the pause and resume on their own — the relay is held down until the archive has finished reloading, so the map pauses rather than running with nothing recording it. Afterwards an installed launcher can tell you when a new release is out, instead of you having to go and look.
Update 2026-08-20 01:32 UTC. Done. The website and the live map returned at 01:28:24 UTC and the pause lasted 1 minute 22 seconds, inside the two minutes announced. Every world that was connected reconnected and reclaimed its slot, and no crossing record was lost — the relay stayed down until the archive had finished reloading and resubscribed. An installed launcher can now tell you when a new release is out.
· Planned maintenance
The website and the live map will be unavailable for up to about four minutes while the archive restarts. Worlds keep running and saving on their own machines throughout; nothing is lost, and no participant action is needed. Crossings pause for the length of the pause and resume on their own. Afterwards the Windows instructions on the front page match the setup the current release actually ships — three steps instead of five, with no menu to learn.
Update 2026-08-18 21:12 UTC. Done, and much shorter than announced. The website and the live map returned at 21:08:17 UTC and the pause lasted 43 seconds. The Windows instructions now read three steps, every world reconnected, and no crossing record was lost.
· Planned maintenance
The website and the live map will be unavailable for about three minutes while the archive restarts. Worlds keep running and saving locally throughout; nothing is lost, and no participant action is needed. Migrations pause for the length of the pause and resume on their own. Afterwards the download section names the release it serves, and it will keep naming the newest one without another maintenance window.
Update 2026-08-18 01:50 UTC. Done, and much shorter than announced. The website and the live map returned at 01:44:59 UTC and the pause lasted 8 seconds. The download section now names the release it serves, and every world reconnected.
· Planned maintenance
Planned maintenance — the website and the live map pause for about six minutes while the homepage is updated and the archive completes a one-time housekeeping step. Worlds keep running and no participant action is needed.
Update 2026-08-17 14:16 UTC. Done. The website and the live map returned at 14:14 UTC and the pause lasted 2 minutes 45 seconds, shorter than the six minutes announced. Worlds kept running throughout.
· Service terms
The map is changing how long it keeps one part of its record, and this notice states exactly which part. The record of what happened is unchanged and is still kept for the whole run and beyond: every species that crossed, how often, when it first and last crossed, and which species it descended from. Nothing removes a species, a count, or a family link.
What changes is the individual crossing lines behind those counts — one line for each organism that crossed. From 2026-08-17 the map keeps those lines for 30 days, the same period it already keeps genome files for. After 30 days a line is no longer on the server. The crossing it recorded is still counted, and the family link it carried is still there.
The operator keeps a compressed copy of those lines away from the server for the length of the run, and a line only leaves the server once that copy is confirmed. That copy is not served from the map and it is not something the map can look things up in.
Why. The lines are about a gigabyte a day. Keeping every one of them on the server to the end of the run needs more space than the server has, and the record — the counts, the dates and the family links — is what the map is actually for. Writing the record down and keeping the lines for a month is what lets the map promise the record for the whole run and mean it.
This is a schedule, not a deletion service. It ages lines by date. It cannot pick out a world, a species or an organism, and it does not change what the operator can do about a name that needs to stop being shown — that is still a change to the view and never to the record.
What the record cannot do. It holds what happened, and nothing more. A crossing that never arrived was never recorded, and there is no way to work one out afterwards from anything the map keeps.
You do not need to do anything. Your world, your saves, your journal and your genomes are on your machine and none of them is affected.
· Planned maintenance
The machine that runs the shared relay, the archive, and this website reboots tonight between 22:30 and 23:30 UTC on 16 August 2026. The service is down for about five minutes inside that window. The reboot applies a kernel update and a fix to how the archive uses memory.
While it is down, the live map and the broadcast page do not answer, and crossings between worlds pause. Your own world keeps running on your own machine throughout. Nothing about it changes.
The crossing record stays complete. The relay is held down until the archive has finished reloading, so the map pauses instead of running with nothing recording it. No crossing is lost from the permanent record.
You do not need to do anything. Each connected world reconnects on its own once the relay is back and reclaims the same slot on the map. Reconnection can take up to about a minute after service returns.
This announcements page comes back before the rest of the site does, so it is the page to check while the map is still loading its record.
A follow-up line below will state the actual outage times once the service is back.
Update 2026-08-16 22:32 UTC. The maintenance planned for tonight did not go ahead. A pre-flight check failed before any part of the service was stopped, so the work was called off rather than started. There was no outage. The live map, the crossing record and this website stayed up for the whole window, and no crossing is missing. The work will be rescheduled, and a new notice will appear here before it happens.
Update 2026-08-16 22:37 UTC. The maintenance has been rescheduled to tonight between 23:00 and 23:45 UTC. The scope is unchanged: a kernel update and a fix to how the archive uses memory, and the service is down for about five minutes inside that window. The relay is held down until the archive has finished reloading, so the crossing record stays complete and no crossing is lost from the permanent record. You do not need to do anything; each connected world reconnects on its own and reclaims the same slot on the map.
Update 2026-08-16 23:04 UTC. The maintenance is complete and the service is back. The outage ran from 23:00:08 to 23:01:51 UTC — 1 minute and 43 seconds, shorter than the five minutes announced. The crossing record is complete for the whole window: the archive was recording again before the first world reclaimed its slot, so nothing that happened on the map went unrecorded. All seven worlds are back on the map and no action is needed from you.