pingoru pingoru Kernel had 5 outages in the last 2 years totaling 12h 11m of downtime, averaging 0.2 incidents per month.
There were 5 Kernel outages since September 10, 2026 totaling 12h 11m of downtime. Each is summarised below. Incident details, duration, and resolution information.
Status: Identified Our captcha solver is not working. The issue has been identified and we are working on triage. Affected components Browsers (Degraded performance)
Status: Resolved This was resolved, apologies for the disruption. Affected components Browsers (Operational)
Status: Identified Between 12:10 and 12:23 PT (19:10–19:23 UTC), a degradation caused most API requests to get, list, update, or delete browsers to fail with 500 errors. Some browser create requests also failed, and browsers that were already running may have had network or proxy connection problems during this window. We rolled back the deploy and error rates returned to normal at 12:23 PT. We're monitoring to confirm everything stays stable. If a browser created during this window is misbehaving, delete it and create a new one. Affected components Browsers (Full outage) Apps (Full outage)
Status: Monitoring As of 12:23pm PT, a fix has been deployed and we are monitoring. Affected components Apps (Full outage) Browsers (Full outage)
Status: Resolved This incident is resolved. The API is healthy again. Affected components Apps (Operational) Browsers (Operational)
Status: Monitoring A browser image released at 14:03 UTC on October 1 causes window.open: http://window.open popups to remain on about:blank while the opening page freezes. Sessions controlled through live view, the computer API, Selenium, or raw CDP are affected. Playwright and Puppeteer sessions are not affected. The cause has been identified and a corrected image is being deployed. We expect it to reach production within the hour. Sessions created after the deployment completes will not be affected. Workaround: if a session is frozen, run the following command against it to resume the popup and the original page. kernel browsers playwright execute 'return 1' We will post another update once the fix is live. Affected components Browsers (Partial outage)
Status: Resolved The fix has been deployed. Affected components Browsers (Operational)
Status: Identified We have been sporadic errors on residential and mobile proxy creation since ~19:00 UTC today. We have traced this to a downstream vendor and are working with them. Existing proxies are healthy and our recommendation is to re-use existing proxies. We will provide updates as we learn more. Affected components Apps (Degraded performance) Browsers (Degraded performance)
Status: Identified The error is coming from increased response latency in our downstream provider. Until this is resolved, we recommend increasing response timeouts. If you are using our SDK, recommended example snippet: import Kernel from '@onkernel/sdk'; // client-wide: applies to every request const client = new Kernel({ timeout: 5 * 60 * 1000, // 5 minutes }); // or per request (options are the last argument) const browser = await client.browsers.create( { stealth: true }, { timeout: 5 * 60 * 1000 }, ); Affected components Apps (Degraded performance) Browsers (Degraded performance)
Status: Resolved Elevated latency affecting residential and mobile proxy creation has subsided, and proxy creation is back to normal. Existing proxy traffic and ISP and datacenter proxy creation were not affected. We’re continuing to monitor. Affected components Apps (Operational) Browsers (Operational)
Status: Investigating There are issues with connecting to some browser instances Affected components Browsers (Partial outage)
Status: Monitoring The issue has stabilized and we are monitoring performance Affected components Browsers (Partial outage)
Status: Resolved Services are operating normally Affected components Browsers (Operational)