Sapling had 18 outages in the last 2 years totaling 217h 33m of downtime — averaging 0.7 incidents per month.
There were 18 Sapling outages since October 18, 2024 totaling 217h 33m of downtime. Each is summarised below — incident details, duration, and resolution information.
We are currently investigating this issue.
We are continuing to investigate this issue.
This incident has been resolved.
From 4:30 to 5:30pm PDT on Sep. 10, 2026, a database migration overloaded the production database with two other read-heavy jobs running at the same time. The issue has been identified and resolved.
Website server needed unexpected reboot
This incident has been resolved.
Upgrade of model resulted in unexpected downtime due to mis-specified model ID
We are continuing to monitor for any further issues.
This incident has been resolved.
From 4:40am to 7:40am PST on Jan 25, 2026, the cluster experienced downtime due to expired trust anchors. New certs have been generated.
We are currently patching this issue -- the extension will not work in Chrome v144
We are continuing to investigate this issue.
A fix has been implemented and we are monitoring the results.
We are continuing to monitor for any further issues.
This has been resolved for over 6 days with the new integration. We left the message so those with the older integration would update.
This incident has been resolved.
AWS cluster upgrades led to insufficient AI detection pod assignment, leading to latency and then downtime starting from 10:30am PST.
Limited resources in AWS region have led to insufficient AI detector pods since 11/5
This incident has been resolved.
We have seen a surge in spellcheck usage in the past day resulting in some scaling of the cluster components needed.
This incident has been resolved.
Subset of pods failing to connect to caching system after caching system upgrade led to degraded services
Delays when writing to S3 resulted in latency spikes across endpoints.
A fix has been implemented and we are monitoring the results.
Increased load to multilingual handling as well as increased latency when sending Amplitude events led to degraded latency for other endpoints.
A shortage of available pods resulted in latency spike starting around 10:20 PDT
Increase in latency due to upgrades to libraries for error classification pipeline.
A fix has been implemented and we are monitoring the results.
This incident has been resolved.
Due to Linkerd issues, services outside of /edits (primarily autocomplete and AI detection) were degraded / unavailable.