Impact
Customers were unable to log into the iSell platform and were shown 503 Service Unavailable errors instead of the usual login page.
Cause & Resolution
The issue was caused by our web servers reaching their maximum processing capacity (known as Max Request Workers). This happened because the servers became overloaded while waiting on responses from the database, which was affected by a separate storage issue. In effect, the web processes stopped responding as a knock-on effect of the database problem.
Our engineering team restored service by manually restarting Apache on the affected servers. The longer-term fix involved resolving the underlying storage issue on the database.
Prevention Measures
We have implemented safeguards to ensure the database cannot fill up again due to replication errors. Additional monitoring and alerting have been introduced so that any database growth or connection saturation is identified and resolved before it can impact the platform.