Major incidents SaaS vendor changes, page 12

Mixpanel reported a major Query API degradation causing HTTP 500 errors and report-loading failures for US Data Residency projects; EU and IN projects were unaffected. It started at 2026-08-26 21:39 UTC and resolved at 2026-08-26 22:45 UTC, lasting 1 h 6 min; a fix was implemented, and no customer data was lost.

Dremio Cloud (US) reported a major outage preventing some classic engines from starting or scaling, from 2026-08-26 21:25 UTC to 2026-08-26 21:55 UTC (30 min). A provisioning artifact was inadvertently removed during a compliance-related infrastructure change; running engines were unaffected. A fix was applied and normal operations confirmed.

Critical Actions and Pages degradation lasted 2 h 49 min, from 2026-08-26 15:11 UTC to 2026-08-26 18:01 UTC. Actions jobs failed to start, then experienced delays exceeding five minutes; Pages performance degraded. Saturated database-primary writes triggered the disruption; failover did not fully restore service. Queues recovered, but 3.7% of early larger-runner jobs remained stuck pending cancellation.

GitHub reported critical Actions and Pages degradation from 2026-08-26 15:11 UTC to 2026-08-26 18:01 UTC (2 h 49 min). Actions jobs failed to start or were delayed; Pages performance degraded. Workflow-trigger database write saturation caused the impact; failover did not fully restore service. At recovery, 3.7% of early larger-runner jobs remained stuck, awaiting cancellation.

Otter.ai reported a major incident delaying Summary and Speaker tagging and also impacting Otter Chat. It started at 2026-08-26 13:46 UTC and lasted 31 h 38 min, resolving at 2026-08-27 21:24 UTC when Otter said its systems were fully restored.

Drata’s major US Public API degradation caused request failures and potentially incorrect framework/control readiness data from 2026-08-26 13:32 UTC to 2026-08-28 23:58 UTC (2 days 10 h). No data was lost; EU and APAC were initially unaffected. Drata deployed fixes and reported normal operation at resolution, with no customer action required.

Smartsheet reported a critical US incident affecting sheet loading, Reports, Dashboards, and premium apps including Dynamic View, WorkApps, Resource Management, and Control Center. It started at 2026-08-26 13:28 UTC and resolved at 2026-08-26 18:23 UTC, lasting 4 h 55 min. Smartsheet implemented a fix and monitored recovery.

Facebook (Meta) syncs failed with “API access disrupted” in a major incident affecting Census Sync Management UI and Sync Engine from 2026-08-26 13:09 UTC to 2026-08-26 14:11 UTC (1 h 2 min). Census identified an Activations Meta app issue and implemented a fix; all affected syncs will recover on their next triggered or scheduled run.

Figma reported a major MCP disruption preventing some Codex and ChatGPT users from seeing tools such as use_figma. The incident started at 2026-08-26 12:08 UTC and resolved at 2026-08-26 16:19 UTC (computed duration: 4 h 10 min); Figma stated impact began at 10:14 UTC and recovery began at 14:03 UTC.

Vercel CLI login (vc login) experienced errors in a major-impact incident from 2026-08-26 03:06 UTC to 2026-08-26 04:28 UTC, lasting 1 h 22 min. The vendor reported identifying the source and testing a fix before marking the incident resolved; affected customers had trouble logging in through the CLI.

Vercel reported a major incident causing login errors with Vercel CLI (`vc login`), preventing affected customers from signing in. It started at 2026-08-26 03:06 UTC and was resolved at 2026-08-26 04:28 UTC, lasting 1 h 22 min. The team reported identifying the source and testing a fix but did not specify the cause.

Buildkite reported a major site-wide outage starting 2026-08-25 22:55 UTC, resolved 2026-08-26 00:18 UTC (1 h 22 min). Internal DNS failure blocked new builds and delayed job dispatch; web, APIs and notifications were affected. Mitigations restored service; no build requests were lost. The postmortem places customer disruption at 22:47 UTC to 23:16 UTC.

Semgrep reported a critical incident affecting Dashboard and Scans and CI/CD, starting 2026-08-25 19:21 UTC and lasting 2 h 40 min. A bad release caused widespread platform failures; rollback restored stability, and a successful release fully resolved the issue at 2026-08-25 22:01 UTC.

WorkOS reported a major Dashboard sign-in outage that could prevent users from authenticating, starting 2026-08-25 17:35 UTC and resolved at 2026-08-25 18:41 UTC, with a computed duration of 1 h 5 min. A fix was applied at 17:41 UTC and confirmed to have resolved elevated sign-in errors at 17:54 UTC.

Anaconda reported a major backend outage affecting APIs, authentication, repositories, EduBlocks and Pyscript, involving an inability to spawn new compute nodes. The recorded start was 2026-08-25 17:10 UTC, though reported impacts began at 15:45 UTC. A fix was implemented; resolution was recorded at 2026-08-25 18:52 UTC, with a computed duration of 1 h 41 min.

Cloudsmith reported a major incident affecting metadata requests and package downloads, starting 2026-08-25 16:38 UTC and lasting 1 h 30 min. Customers experienced latency and timeouts between 16:04 and 17:13 UTC, with a recurrence between 18:48 and 19:08 UTC; Cloudsmith confirmed normal operations at 08-26 07:55 UTC.

Pinecone reported a major incident affecting Serverless Indexes in GCP us-central1, with 5xx errors on some read and write requests. It started at 2026-08-25 15:50 UTC and was resolved at 2026-08-25 16:11 UTC, lasting 21 min; Pinecone stated that a fix was implemented.

Confluence reported a major incident delaying email, mobile and slack notifications for Jira, JSM and Confluence, starting 2026-08-25 08:42 UTC and resolved at 2026-08-25 14:33 UTC, lasting 5 h 51 min. Teams applied corrective measures in one impacted region; the vendor confirmed normal service for all customers without specifying a cause.

Jira reported a major degradation delaying email, mobile and slack notifications for Jira, JSM and Confluence from 2026-08-25 08:42 UTC to 2026-08-25 14:33 UTC, lasting 5 h 51 min. Teams applied corrective measures in an impacted region; the vendor confirmed resolution and normal service for all customers.

BigCommerce reported a major Control Panel incident affecting Microsoft Azure AD Single Sign-On login for a subset of stores. It started at 2026-08-25 06:56 UTC and was resolved at 2026-08-25 07:25 UTC, lasting 29 min; BigCommerce identified the issue and reported implementing a fix before resolution.

Box reported a critical download issue affecting Box Zones users in the Asian region, who may have been unable to download files. It started at 2026-08-25 06:53 UTC and was resolved at 2026-08-25 07:37 UTC, lasting 44 min; Box reported improvement after remediation steps, and its postmortem also noted ZIP downloads were affected.

Confluent reported a major Confluent Cloud degradation in Azure Southeast Asia, with elevated errors, latency and intermittent connectivity affecting produce/consume availability. Recorded from 2026-08-25 04:17 UTC to 2026-08-25 07:29 UTC (3 h 11 min), it reportedly began around 03:13 UTC and was associated with a Microsoft Azure disruption; services returned to normal.

Snowflake reported a critical outage in Azure - Southeast Asia (Singapore), where infrastructure connectivity issues may have prevented customers from accessing services. It started at 2026-08-25 03:53 UTC and was resolved at 2026-08-25 04:35 UTC, lasting 41 min; Snowflake implemented a fix and confirmed restoration through monitoring. Other regions were unaffected.

Box reported a major-impact upload incident starting 2026-08-24 21:21 UTC, resolved at 2026-08-24 22:03 UTC (41 min). Users could encounter upload errors or slowness across the Box web application and API; the postmortem also noted affected write operations. Box reported remediation and improvement but did not specify the cause or fix in the available evidence.

Descript reported a major Share Page outage with a recorded start of 2026-08-24 18:56 UTC and resolution of 2026-08-24 17:15 UTC; duration is unknown. The vendor said a fix had been implemented and the issue was resolved.

DigitalOcean reported a critical Cloud Control Panel/Public API outage starting 2026-08-24 17:14 UTC, resolved 2026-08-25 19:04 UTC (25 h 49 min). Customers faced API errors and panel access failures; mitigation restored access. The postmortem reports 3.5 hours of active degradation, peaking at 25% errors; running Droplets, network routing and Volumes remained functional.

Snowflake reported a major Snowsight authentication outage in AWS - US West (Oregon), causing immediate logout after successful authentication. The vendor stated a start of 15:00 UTC August 24, 2026; recorded duration was 1 h 41 min, resolved at 2026-08-24 18:03 UTC. Snowflake attributed the issue to a system configuration and implemented a fix.

Paddle reported a major Retain outage, recorded from 2026-08-24 08:35 UTC to 2026-08-24 09:22 UTC (47 min), though its initial update said the issue began around 7:56 UTC. A fix was deployed at 09:11 UTC, and all services were operating normally at resolution.

GitHub reported a major degradation affecting Claude Fable 5 in Copilot products and IDE surfaces, with elevated errors and requests failing mid-response; no other models were impacted. The incident started at 2026-08-24 07:12 UTC and was resolved at 2026-08-24 07:58 UTC, lasting 46 min. GitHub attributed it to an upstream provider issue, resolved through the provider’s mitigation.

GitHub Copilot reported a major degradation affecting Claude Fable 5, with elevated errors and requests failing mid-response; no other models were impacted. The incident started at 2026-08-24 07:12 UTC and was resolved at 2026-08-24 07:58 UTC, lasting 46 min. An upstream provider issue caused the disruption, and the provider applied a mitigation.

Claude reported a major incident involving elevated errors across claude.ai, Claude API (api.anthropic.com), Claude Code and Claude Cowork, affecting Claude Mythos 5, Claude Fable 5, Claude Opus 5 and Claude Opus 4.8. It started 2026-08-24 05:06 UTC and was resolved 2026-08-24 08:30 UTC, lasting 3 h 24 min.

Anthropic reported a major incident involving elevated errors on claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork. Started: 2026-08-24 05:06 UTC; resolved: 2026-08-24 08:30 UTC; duration: 3 h 24 min. The final update reported user errors from 04:50 UTC through 07:36 UTC, including requests to Claude Opus 5 and Fable 5.

Fly.io reported a major network incident in the Los Angeles (LAX) region, where apps could experience higher latency or be unreachable. It started at 2026-08-23 01:28 UTC and was resolved at 2026-08-23 02:10 UTC, lasting 41 min. Fly.io stated that upstream networking issues had resolved.

A critical incident caused 500 errors when accessing Lever, affecting Global Data Center - LeverTRM - Hire and Hire. It started at 2026-08-21 20:12 UTC and was resolved at 2026-08-24 14:16 UTC, lasting 2 days 18 h. Lever implemented a fix at 08-21 20:29 UTC and promised an RCA within 7-10 business days.

Cloudsmith reported a major global package-caching outage affecting Package Downloaded webhooks; package pulls remained unaffected. Recorded from 2026-08-21 18:15 UTC to 2026-08-21 18:54 UTC (39 min), impact reportedly began approximately 17:00 UTC. Caching resumed after a fix; packages pulled between 17:00 and 18:45 UTC were missing from repositories and will be cached on subsequent pulls.

Lattice reported a major incident affecting the admin Compensation Cycle Recommendations table, which showed a "Something Went Wrong" error, starting 2026-08-21 16:30 UTC. It was resolved at 2026-08-21 18:24 UTC, lasting 1 h 53 min. A fix was deployed; admins were instructed to refresh the page to see data successfully.

Drata reported critical SafeBase request failures: some requests errored or timed out because edge traffic filtering rules blocked legitimate traffic. Recorded start and resolution were both 2026-08-21 01:00 UTC (computed duration: 0 min), although the vendor described approximately six minutes of failures on August 20.

Socket reported a major Socket API incident in which the Events collection system incorrectly rejected Firewall Events. It started at 2026-08-20 23:05 UTC and was resolved at 2026-08-21 01:06 UTC, lasting 2 h 1 min; customers’ Firewall Events could be rejected or dropped during that period.

Claude reported a major incident involving elevated errors on requests to multiple models, affecting claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork. It started at 2026-08-20 19:16 UTC and was resolved at 2026-08-20 19:42 UTC, lasting 26 min; no cause or fix details were stated.

Anthropic reported a major incident involving elevated errors on requests to some Claude models, affecting claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork. It started at 2026-08-20 19:16 UTC and was resolved at 2026-08-20 19:42 UTC, lasting 26 min; no cause or specific fix was stated.

Ably reported a major incident with elevated errors in ap-southeast-1, affecting Ably Pub/Sub - Asia-Pacific (Singapore) and Asia-Pacific (Singapore). It started at 2026-08-20 19:07 UTC and was resolved at 2026-08-20 19:10 UTC; the computed duration was 2 min. No cause or fix was stated.

v0 reported a major Functions incident: newly deployed container functions could return 5XX errors starting at 2026-08-20 17:46 UTC. Customers could temporarily use a previously working deployment; a fix was deployed at 2026-08-20 18:23 UTC. The incident was resolved at 2026-08-20 18:34 UTC, lasting 47 min.

Vercel reported a major Functions outage: newly deployed container functions could return 5XX errors from 2026-08-20 17:46 UTC until resolution at 2026-08-20 18:34 UTC (47 min). Customers could temporarily use a previously working deployment; Vercel reported deploying a fix at 18:23 UTC, restoring newly deployed container functions.