Server record
How steadily the NanumOS server stays open, recorded as it is. More than a year of real operation is the proof that the server can be trusted, and this page is that record. Bad moments are not hidden.
Checked every minute from 2 places. Numbers as of 2026-11-16 09:03 (KST).
The 30 / 90 / 180 / 365 day steps
The first 30, 90, 180 days and 1 year of the record fill in one by one. Once a step is reached its numbers are fixed.
| Step | State | Uptime | p95 | Incidents |
|---|---|---|---|---|
| 30 days | in progress (so far) | 98.976% | 166 ms | 1 |
| 90 days | in progress (so far) | 98.976% | 166 ms | 1 |
| 180 days | in progress (so far) | 98.976% | 166 ms | 1 |
| 365 days | in progress (so far) | 98.976% | 166 ms | 1 |
Which server answered, day by day
Of the checks that came in through the nanumos.com address, the share the NanumOS server answered itself. Days below 100% were answered by the old server; the day of the move shows here.
Incidents
Two or more failed checks in a row count as one incident. A cause is written only when known; otherwise it says so.
- 2026-11-16 00:54 (KST) · 16.2 min · Every checking place saw it. The cause has not been written up yet.
On the nanumos.com address (the way everybody comes in): 0 incidents so far, uptime over the last 30 days 99.520%.
The security certificate is valid for 82 more days.
How the numbers are made
Each of 2 different places (outside the server, on different networks) fetches the front page once a minute and notes the status, the time to the first byte and to the end, which server answered, and a fingerprint of the page.
Technical details
- A good check is a 2xx/3xx answer within 15 seconds. Uptime = good checks / all checks over all places, the strict reading: one place's bad network costs points.
- p50 and p95 are the median and the 95th percentile (nearest rank) of the total time of good checks.
- An incident is two or more failures in a row from the same place for the same address; overlapping ones from several places merge into one. A gap in the records (a broken checker) is not counted as a failure.
- A day is a Korea-time day. A step's numbers are fixed from the start of the record to that many days.
- The numbers are recounted every day from the raw records, which are append-only files kept at each checking place.
- Share of checks where the server's front page was identical to the one everybody gets (last 7 days): 20.2%. Low before the move day is normal.
NanumOS 0.1.8349 · 2026-11-16