Evidence: Buffer, Nov 7, 2026

No failed checks with an error code in this window.

http_5xx = the site answered with that server error; timeout = no answer in time; network = the connection failed; origin_reset = the connection to the site's server was reset (HTTP 520–527) and a second fetch path got no answer either; no_status = no usable answer. Bot walls (bot_*), certificate problems on a server that answers (cert_*) and maintenance pages count as up.

• We check from: US East, US West, Europe, Asia-Pacific.
• Failing during this incident: none recorded.

Regional probes run from Cloudflare data centers in each region. We keep the latest regional result per region, plus a timestamped timeline entry when a region starts failing during an incident, so older incidents show the failing regions and when each was first seen, not every regional probe.

No user reports for Buffer in this window.

User reports per 15-minute bucket. Counts only: no reporter identity, IP address, location or text is published. We call reports a spike at 5 or more in 15 minutes.

224 checks, 0 failed.

Every main HTTP check of this service in the window, oldest first: the scheduled check time, when the probe actually ran, the result, HTTP status, latency and the stored error code. A failed check is a 5xx or 408 answer, a timeout or a failed connection; bot walls and maintenance pages count as passing and are noted in passing_notes.

• Compare checks.rows with the outage page: the first and last failed checks bound the incident times we published.
• error_codes counts every failed check by its stored code; recount it from checks.rows.
• report_curve.points sum to summary.reports_in_curve; the peak is summary.peak_reports_15m.
• Disagree? Dispute it at https://damnisitdown.com/support?dispute=1066

No personal data: no reporter identities, IP addresses, emails, report text or reporter locations. User reports are aggregate 15-minute counts.

← Back to the incident · Methodology · Dispute this report