Lever Incident History

Issue: We identified a platform-wide issue in which offer letters failed to be sent. Customer Impact: Recruiters were unable to generate or deliver offer emails to candidates. Status: Resolved. An RCA will be available within 7-10 business days.

Issue: We identified a platform-wide issue in which offer letters failed to be sent. Customer Impact: Recruiters were unable to generate or deliver offer emails to candidates. Current Status: The immediate issue has been mitigated. We are actively monitoring to confirm full resolution and stability.

Issue: We are investigating what appears to be a platform-wide issue in which offer letters fail to be sent. Customer Impact: Recruiters are currently unable to generate or deliver offer emails to candidates. Current Status: Engineering is actively investigating. No workaround is available at this time.

Issue Description: Users may have received an "Unknown error" when attempting to send emails or sync their Microsoft email and calendar in Lever. Customer Impact: Affected users were unable to authenticate their Microsoft email and calendar integrations, blocking email sends to candidates and calendar sync functionality critical to recruiting workflows. Status: Resolved — Email sends and calendar sync are now functioning normally. An RCA will be available within 7-10 business days.

Mitigation steps have been applied and service functionality is returning to normal. We are monitoring performance and stability. Next update: upon resolution.

We are continuing to investigate this issue. Next update: 5:00 PM PT

We are continuing to investigate this issue. Next update: 3:00 pm PT or sooner.

We are investigating an issue affecting Nylas authentication. Users may experience an error message "Unknown error" when attempting to reset email/calendar integration permissions. Start time: 10:10 am PT Next update: 1:00 pm PT or sooner.

**Incident Summary** Between July 2, 2026, and July 8, 2026, Some customers experienced intermittent failures affecting email and calendar services. The primary impacts included errors when sending or syncing emails and occasional "500 errors" during login. These issues were caused by a technical regression in the underlying runtime environment \(Node.js\) that handles outbound connections to third-party services. For a subset of customers using specific email integration settings, the connection instability caused the system to incorrectly perceive a successful email send as a failure. This triggered an automated retry mechanism, which in some cases resulted in a "retry loop" where the same communication was re-sent to recipients at regular intervals until the connection was stabilized. **Detection** The issue was identified following reports of intermittent service disruptions. Our support and engineering teams conducted a thorough investigation into these reports, identifying a specific pattern of connection failures within our system logs. Further analysis confirmed that the impact began on approximately July 2, 2026, following a routine update to our service infrastructure. **Root Cause** The root cause was a defect in a specific version of Node.js \(version 24.17.0\), the software environment used to run our services. This version contained a flaw in how it managed "keep-alive" connections—reusable connections used to communicate with external services. Under certain conditions, the software would prematurely close these connections before a task was completed, resulting in failed email sends and data sync errors. **Resolution** To resolve the issue, our engineering team performed the following steps: * **Identification:** Correlated the "Premature close" errors to a known regression in the Node.js runtime. * **Upgrade:** Updated our base system images to Node.js version 24.18.0, which contains the official fix for this connection issue. * **Deployment:** Rebuilt and redeployed critical services \(including email and calendar components\) on the corrected version. * **Verification:** Confirmed that connection errors ceased and service stability returned to normal levels on July 8, 2026. **Preventative Measures** To prevent similar issues in the future, we are implementing the following improvements: * **Version Pinning:** We have updated our build process to ensure we deploy only versions that have been fully verified. * **Enhanced Monitoring:** We are adding specific alerts for outbound connection errors to detect and respond to similar communication failures more rapidly. * **Automated Testing:** We are implementing new synthetic checks for email and calendar synchronization to proactively identify issues before they impact customers. * **Improved Visibility:** We are updating our internal dashboards to provide better visibility into the exact software versions running in our production environment to speed up future diagnoses.

This incident is resolved. A Root Cause Analysis (RCA) will be published to this issue within 7-10 business days.

A fix has been implemented as of approximately 12:15pm PST/PDT on 7/8/2026. Email and Calendar sync are now fully functional for Lever instances configured with Nylas authentication. We are continuing to monitor the effectiveness of the fix. A Root Cause Analysis (RCA) will be published to this issue within the standard Service Level Agreement.

On 7-3-2026 we were made aware of an issue impacting Lever instances with Nylas authentication (generally used for customers with on-premise Microsoft Exchange email/calendar services). Only Nylas authentication is impacted, there has been no similar behavior observed for customers authenticating with Office 365 or Google services. Users may experience a "500 error" message when attempting to reset email/calendar permissions and we have observed several unexpected email loops causing candidates to receive duplicate emails every 10 to 15 minutes as a result of this issue. Root cause: The 500s and email failures are caused by a known bug in Node.js v24.17.0 (https://github.com/nodejs/node/issues/63989). The v24.17.0 security release (CVE-2026-48931) changed `http.Agent` keep-alive socket handling so that `node-fetch` throws `ERR_STREAM_PREMATURE_CLOSE` ("Premature close") on gzipped responses to external providers (Nylas/Google). Our services picked it up this bug via the Node 18→24 migration (Occurred on 06-30-2026). The bug was fixed upstream in Node 24.18.0 This issue should be fixed in a deployment today (7-8-2026). We will provide updates on the issue status and deployment timeline as soon as possible.

This incident is resolved. A Root Cause Analysis (RCA) will be published to this issue within 7-10 business days.

A fix has been implemented as of approximately 12:15pm PST/PDT on 7/8/2026. Email and Calendar sync are now fully functional for Lever instances configured with Nylas authentication.We are continuing to monitor the effectiveness of the fix. A Root Cause Analysis (RCA) will be published to this issue within the standard Service Level Agreement.

On 7-3-2026 we were made aware of an issue impacting Lever instances with Nylas authentication (generally used for customers with on-premise Microsoft Exchange email/calendar services). Only Nylas authentication is impacted, there has been no similar behavior observed for customers authenticating with Office 365 or Google services.Users may experience a "500 error" message when attempting to reset email/calendar permissions and we have observed several unexpected email loops causing candidates to receive duplicate emails every 10 to 15 minutes as a result of this issue.Root cause: The 500s and email failures are caused by a known bug in Node.js v24.17.0 (https://github.com/nodejs/node/issues/63989). The v24.17.0 security release (CVE-2026-48931) changed `http.Agent` keep-alive socket handling so that `node-fetch` throws `ERR_STREAM_PREMATURE_CLOSE` ("Premature close") on gzipped responses to external providers (Nylas/Google). Our services picked it up this bug via the Node 18→24 migration (Occurred on 06-30-2026). The bug was fixed upstream in Node 24.18.0This issue should be fixed in a deployment today (7-8-2026). We will provide updates on the issue status and deployment timeline as soon as possible.