Is there a backup of FogCam?

Yes. Fog-Bank (fog-bank.org) backs up every frame FogCam publishes: four independent pollers on three networks fetch it within seconds, it is sha-256 verified, converted losslessly to JPEG XL and kept on the archive server, on a Cloudflare R2 mirror, in a 30-day off-site buffer and on a home machine.

How often is FogCam backed up?

Continuously. Four pollers on three networks ask the camera for its image every 7 to 20 seconds, so every frame it publishes (one every 20 to 60 seconds) is captured. Fog-Bank holds more than 262,000 frames and adds each new one as it appears.

What happens when FogCam goes down?

The gap is recorded with its cause. If the camera stops publishing, the record says so, and nothing was lost that was never published. If the archive's own capture fails, that is recorded as Fog-Bank's fault, an alert goes out, and the second poller's buffer fills the gap.

How FogCam is backed up

Fog-Bank, the FogCam archive · fog-bank.org

FogCam keeps one picture and overwrites it with the next. Fog-Bank is the backup: every frame the fog cam publishes gets fetched within seconds by four pollers on three networks and kept in four places. This page is the plain description of how, with the numbers it produces (they're live, hover one to see how it was made).

Capture

A poller on my server asks fogcam.org for the image every 7 seconds. Each request carries a unique query string and no-cache headers so FogCam's own server cache can't hand back an old copy, plus a conditional header so an unchanged frame costs almost nothing to check. A new frame is named by the camera's own Last-Modified time and hashed with sha-256 before anything else touches it. Every attempt (new, unchanged or failed) goes into a permanent ledger.

A second poller, off-site

fogline runs on Cloudflare's network and polls the camera every 15 seconds over a different path. It keeps the last 30 days of raw frames. Every frame it gets is pushed straight into the archive, and a sweep every 5 minutes and every night pulls anything that was missed, so a reboot, a full disk or a dead link on my side loses nothing fogline saw. It is also the witness: when my side goes quiet and fogline is still getting frames, the fault is mine and I get a Telegram message saying so.

A third and a fourth poller

A worker on Cloudflare polls the camera every 20 seconds on its own and writes each new frame straight into permanent storage (R2). A fourth poller runs on my desktop at home, in another state. It saves every frame to its own disk first (30 days) and sends it to the archive and the mirror within a minute, so a Cloudflare or hosting outage costs it nothing (it only runs while the desktop is on).

A mirror

Every frame from every poller gets copied to Cloudflare R2 with its hash and indexed by day, and the archive checks itself against the mirror every 5 minutes and every night. The older archive, the 2019 rescue included, went up in September 2026, and the database (the ledger, incidents and adoptions) gets copied there every night too. Losing a frame now takes all four pollers missing the same picture at once, on a camera that holds each one for at least 20 seconds.

Storage

Frames are converted to lossless JPEG XL within minutes. It converts back bit for bit, the original sha-256 is kept, and a frame served from the site is the original JPEG rebuilt. The 27 damaged files in the 2015 and 2019 rescue ZIPs are kept out of the count, not repaired or passed off as frames.

Each layer watched

An off-site check runs every 5 minutes. It reads the archive's status from outside, compares it with what it's catching itself, and checks the other pollers' pulse. A status board (on the net) checks every stage from the camera to YouTube, and a stage that stops gets reported on Telegram while the others keep catching.

Gaps, and whose they are

When a day has a drop, the record says which side caused it. My poller silent: mine. FogCam answering with the same picture for half an hour or more, or not answering at all: FogCam's. Both are drawn on the calendar and on every chart with a time axis, and a drop of mine pages me. The full story, including the day I found an upstream cache taking frames, is on the net.

drop: my capture was downdrop: FogCam published nothingunsure: connection errors while the camera was quiet

The numbers

262,968 frames held across 422 days, the oldest from 2005-05-18
128 complete days
FogCam down 229.8 h in total since the line began, my capture down 82.6 h
113 quiet stretches (3.7 h in all) where my server had connection errors to the camera, so a frame could have been published unseen. Those are the only stretches the archive can't vouch for (amber on every chart)
Today: 1,115 frames held of 1,116 FogCam published, as witnessed by fogline
126,835 frames from the 2019 community rescue

Code and configuration

The archive's code and settings get bundled every hour to an off-site server, so the whole thing can be rebuilt on a clean machine.

More: the net, in full · the frames that were lost before · browse the archive by date · why the archive exists