Notable SaaS vendor changes, page 14

GitHub reported a major Actions outage starting 2026-08-18 09:36 UTC, resolved at 10:23 UTC (46 min). Its update separately cited 05:02 UTC–11:30 UTC: customers could not run Actions Larger Runners jobs or view/manage Actions Runners and Runner Groups through the GitHub UI/API. The vendor attributed this to failed backend requests resolving essential metadata.

ElevenLabs reported a major degradation affecting <1% of inbound Twilio calls, preventing agent connections; outbound calls and other traffic were unaffected. The incident ran from 2026-08-18 06:00 UTC to 2026-08-18 12:43 UTC (6 h 43 min). A configuration error was rolled back; calls connected normally since ~11:00 UTC, with no customer action needed.

Kimi K3 experienced elevated errors and degraded availability during a major incident starting 2026-08-18 05:52 UTC, caused by connectivity issues affecting one region. Availability recovered by 06:22 UTC; the incident was resolved at 2026-08-18 07:01 UTC after 1 h 9 min.

Astro Hosted experienced intermittent task instance failures with major impact from 2026-08-18 04:17 UTC to 2026-08-18 07:36 UTC, lasting 3 h 18 min. Astronomer identified an issue affecting an internal rollout related to the pre_execute Airflow hook and resolved the incident after completing a fix rollout across all Astro deployments.

SurveyMonkey reported a major outage: all Summary and Responses Data exports to Excel/CSV/SPSS failed, while PDF and PPT exports still worked. It started at 2026-08-18 00:42 UTC and was resolved at 2026-08-18 01:03 UTC, lasting 20 min; a fix was implemented.

Matillion reported a critical disruption that made projects inaccessible across all regions, affecting Designer and Pipeline Execution in AU, EU and US regions. It started at 2026-08-17 17:53 UTC and lasted 55 min, resolving at 2026-08-17 18:48 UTC; a fix was implemented, but no cause was stated.

Matillion reported a critical AU Region outage that made projects inaccessible, affecting Designer and Pipeline Execution. It started at 2026-08-17 17:40 UTC and was resolved at 2026-08-17 17:42 UTC; the computed duration was 1 min. No cause or specific fix was stated.

Docker reported a major incident affecting builds of DHI customizations from 2026-08-17 15:21 UTC to 2026-08-17 22:04 UTC, lasting 6 h 43 min. Docker attributed the disruption to a third-party vendor outage and confirmed customizations were going through after the GitHub outage ended.

dbt Labs reported a major Git outage starting 2026-08-17 15:02 UTC, lasting 6 h 19 min and resolving at 2026-08-17 21:21 UTC. A widespread GitHub outage caused failures in dbt deps, repository connections, Studio IDE, Scheduled Jobs and GitHub authentication; dbt confirmed all affected systems were functioning normally.

CircleCI reported major degradation affecting pipeline triggering and login, with possible failures reporting running jobs’ status to GitHub Pull Requests. The incident ran from 2026-08-17 14:04 UTC to 2026-08-17 19:05 UTC, lasting 5 h 1 min. CircleCI attributed the disruption to a GitHub incident and resolved it after GitHub’s APIs appeared to be operating normally.

GitHub reported a critical outage affecting Issues, Pull Requests, APIs, Actions and Copilot, with elevated errors and latency. Peak web/API error rates were approximately 20%; archive and raw-content downloads reached approximately 50%. It started 2026-08-17 13:40 UTC and was resolved at 2026-08-17 21:15 UTC, lasting 7 h 35 min.

GitHub reported a critical outage affecting Copilot authentication and multiple GitHub.com services, with elevated errors and latency. It started at 2026-08-17 13:40 UTC and resolved at 2026-08-17 21:15 UTC (7 h 35 min); mitigations addressed sporadic Copilot authentication failures, while Copilot usage via the GitHub CLI and GitHub App remained unaffected.

Convex reported a major outage preventing dashboard emails, including team invitations and new account verification, across Business, Free & Starter, and Professional. It started at 2026-08-16 23:33 UTC and lasted 1 h 53 min, resolving at 2026-08-17 01:26 UTC; a fix was implemented at 2026-08-17 00:50 UTC.

Claude reported a critical disruption involving authentication issues on claude.ai, Claude Code, and Claude Cowork, plus degraded performance across those services, Claude Console, and the Claude API. It started at 2026-08-16 21:58 UTC and was resolved at 2026-08-16 22:34 UTC, lasting 35 min; a fix was deployed at 22:22 UTC.

Anthropic reported a critical disruption involving authentication issues and degraded performance across claude.ai, Claude API, Claude Code, Claude Console, and Claude Cowork. It started at 2026-08-16 21:58 UTC and was resolved at 2026-08-16 22:34 UTC, lasting 35 min; a fix was deployed at 22:22 UTC.

Brex reported a critical Mastercard Network outage affecting card authorization, causing transactions to be incorrectly declined. It started at 2026-08-15 04:49 UTC and was resolved at 2026-08-15 05:53 UTC, lasting 1 h 4 min; Brex said transactions should now be correctly processed.

Claude Fable 5 requests on claude.ai experienced elevated errors, rated major. The incident started at 2026-08-14 23:57 UTC and resolved at 2026-08-15 00:27 UTC (30 min). Anthropic separately reported impact from 20:00 Aug 14, 2026 to 00:11 Aug 15 , 2026 UTC; no specific cause or fix was disclosed.

Claude reported elevated errors for Claude Fable 5 affecting claude.ai, with major impact. The incident started at 2026-08-14 23:57 UTC and was resolved at 2026-08-15 00:27 UTC (30 min). The resolution update separately stated customer impact occurred from 20:00 Aug 14, 2026 to 00:11 Aug 15, 2026 UTC.

RunPod’s US-IL-1 datacenter was unavailable due to a power outage, with critical impact, from 2026-08-14 21:30 UTC to 2026-08-15 03:40 UTC, lasting 6 h 10 min. Power was restored to most servers by 08-15 02:59 UTC; the vendor confirmed normal datacenter operation at resolution.

A major degradation affected Claude API (api.anthropic.com), Claude Code, and Claude Cowork. The incident started at 2026-08-14 20:30 UTC and was resolved at 2026-08-14 21:58 UTC (1 h 27 min); Anthropic reported customer impact from 20:14 to 20:38 UTC. A fix was deployed at 20:44 UTC, but no specific cause was disclosed.

Claude reported a major degradation affecting Claude API (api.anthropic.com), Claude Code, and Claude Cowork, starting 2026-08-14 20:30 UTC and resolved at 2026-08-14 21:58 UTC (computed duration: 1 h 27 min). The vendor reported actual impact from 20:14 to 20:38 UTC. A fix was deployed at 20:44 UTC; the cause was not specified.

Fly.io’s major secrets service outage blocked new app creation and secret updates; existing apps continued running. It lasted 1 h 3 min, from 2026-08-14 19:30 UTC to 2026-08-14 20:33 UTC. Fly.io cited a failed deployment and failed over the secrets database to a replica, after which the Machines API appeared healthy.

Snowflake reported a critical Snowsight outage in Azure - East US 2 (Virginia), blocking access/sign-in with “503 Service Unavailable” errors. Recorded from 2026-08-14 17:29 UTC to 2026-08-14 19:35 UTC (2 h 5 min), customer impact reportedly began at 17:05 UTC August 14, 2026. Snowflake implemented a fix for a system configuration issue and confirmed service restoration.

Unbounce reported a major Page Servers outage in Asia Pacific: visitors to live pages could see gateway errors; no other regions were affected. It started at 2026-08-14 15:28 UTC and was resolved at 2026-08-14 17:42 UTC, lasting 2 h 14 min. Unbounce implemented a fix and reported health checks back to 100%.

Grammarly reported a major outage involving login issues across some Grammarly and Superhuman products and preventing some users from accessing Grammarly Editor documents. It lasted 2 h 33 min, from 2026-08-14 15:24 UTC to 2026-08-14 17:57 UTC; a fix was implemented and the incident was resolved.

Snowflake reported a critical Snowsight availability outage affecting Applications in AWS - US East (Ohio), with intermittent authentication failures and "service unavailable" errors. The incident started at 2026-08-14 15:19 UTC and resolved at 2026-08-14 16:28 UTC; recorded duration was 1 h 8 min. Automated recovery restored service; a postmortem links to an RCA on Snowflake Community.

Retool reported a major incident causing errors when calling resources with Per-User authentication in Apps built with the New App Editor. It started at 2026-08-14 15:06 UTC and was resolved at 2026-08-14 19:00 UTC, lasting 3 h 53 min. A fix was implemented by 2026-08-14 16:36 UTC; no cause was stated.

Sanity reported a major incident affecting api.sanity.io and apicdn.sanity.io, with elevated error rates identified from 14:34 UTC until a fix restored normal rates at 14:40 UTC. The incident record started at 2026-08-14 14:46 UTC and was resolved at 2026-08-14 15:32 UTC; its recorded duration was 45 min.

Elastic Cloud reported a major Elasticsearch 9.5.1 issue causing false-positive matches in boolean queries combining multi-value terms in must, filter, or should clauses with must_not on fields with disabled indexing. It ran from 2026-08-14 09:22 UTC to 2026-08-20 17:09 UTC (6 days 7 h); Elasticsearch 9.5.2 fixes it, and customers on 9.5.0 or 9.5.1 should upgrade.

Discord reported a major incident affecting call establishment from 2026-08-14 06:09 UTC to 2026-08-14 06:24 UTC, lasting 14 min. A fix was implemented by 2026-08-14 06:14 UTC, and the incident was resolved.

Sentry reported major ingestion delays for spans and transactions in DE and US, with a US errors backlog also noted. The incident started at 2026-08-13 17:47 UTC and lasted 4 h 39 min, delaying ingestion of affected customer telemetry. Sentry resolved it at 2026-08-13 22:27 UTC after all backlogs were processed.

Grafana Cloud reported a critical K6 Test Outage starting 2026-08-13 15:25 UTC, lasting 1 h 22 min and resolved at 2026-08-13 16:47 UTC. Telemetry push APIs went down at 15:06 UTC, causing tests to lose metrics, logs, traces and browser screenshots; Grafana reported mitigation but did not state a root cause.

Claude reported a major incident involving elevated request errors for Claude Mythos 5, Claude Fable 5, and Claude Sonnet 5, affecting claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork. It started at 2026-08-13 14:33 UTC and was resolved at 2026-08-13 16:08 UTC, lasting 1 h 34 min; the resolution update named Mythos 5 and Fable 5.

Anthropic reported a major incident with elevated request errors for Claude Mythos 5, Claude Fable 5, and Claude Sonnet 5, affecting claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork. It started at 2026-08-13 14:33 UTC and was resolved at 2026-08-13 16:08 UTC, lasting 1 h 34 min; no cause or fix was stated.

Contentful reported a major incident causing Content Management API latency and errors for some customers in the EU data residency region. It started at 2026-08-13 09:34 UTC and was resolved at 2026-08-13 11:38 UTC, lasting 2 h 4 min; Contentful said the issue was addressed without specifying a cause or fix.

Elastic Cloud reported a major incident starting 2026-08-13 08:32 UTC, lasting 4 days 3 h, with search embedding failures for microsoft-multilingual-e5-large due to upstream provider issues. Resolved 2026-08-17 11:40 UTC, the model was retired from Elastic Inference Service; requests now return errors, and customers must create custom inference endpoints using their own API keys.

Elastic Cloud reported a major outage delaying or failing Serverless project creation and settings changes through the Cloud UI or API; running projects continued operating normally. It started at 2026-08-13 03:57 UTC and was resolved at 2026-08-13 08:40 UTC, lasting 4 h 42 min. A fix was deployed by 06:04 UTC; no cause was stated.

Perplexity reported a critical API issue starting at 2026-08-13 00:45 UTC, resolved at 2026-08-13 01:49 UTC after 1 h 4 min. The vendor identified a problematic deploy and rolled back the service; the incident affected API service, but specific customer-facing symptoms were not stated.

Acuity Scheduling had a critical service disruption starting 2026-08-12 22:55 UTC; Squarespace said all users and services may be affected. Engineering identified the issue, and the incident was resolved at 2026-08-12 23:12 UTC after 16 min; no specific cause or fix was stated.

Sentry reported a major incident affecting US Error Ingestion and US Errors Alerting from 2026-08-12 21:58 UTC to 2026-08-12 22:52 UTC, lasting 54 min. Error ingestion was delayed; Sentry mitigated the source of latency and processed the backlog before resolving the incident.

GitHub reported a major disruption affecting login pages and Release Asset downloads, starting 2026-08-12 21:39 UTC and resolved at 2026-08-12 22:56 UTC (1 h 17 min). Some logged-out requests received HTTP 5xx errors because an unusual traffic pattern repeatedly overloaded infrastructure serving those requests.

GitHub reported a major disruption affecting login and Release Asset downloads, starting 2026-08-12 21:39 UTC and resolved 2026-08-12 22:56 UTC (1 h 17 min). Some logged-out requests received HTTP 5xx errors; GitHub attributed these to an unusual traffic pattern repeatedly overloading infrastructure serving those requests.

Metabase Store was unavailable from 2026-08-12 19:13 UTC to 2026-08-12 20:52 UTC, a critical outage lasting 1 h 39 min. Metabase reported maintenance on the Store infrastructure and subsequently marked the incident resolved.

Metabase Store was unavailable with critical impact from 2026-08-12 19:13 UTC to 2026-08-12 20:52 UTC, lasting 1 h 39 min. Metabase reported maintenance on the Store infrastructure and initially expected it to be operational within an hour; the incident was subsequently resolved.

Expo reported a major outage causing login errors in Expo Go and Dev Clients, starting 2026-08-12 19:00 UTC and resolved at 2026-08-13 00:00 UTC, lasting 5 h 0 min. Expo confirmed a fix was deployed; no cause was stated.

Semgrep reported a major incident affecting Managed Scans: lockfileless scans failed for certain deployments from 2026-08-12 18:16 UTC until resolution at 2026-08-12 21:03 UTC, lasting 2 h 46 min. A fix was rolled out by 2026-08-12 20:23 UTC, restoring scans; no specific root cause was disclosed.