Incident resolved. All searches should be working normally again. Recent sessions may take some time to appear as the queue of recordings is processed.
Posted Nov 22, 2016 - 19:58 EST
Update
Recent increases in application errors such as "Failed to load users" correlate with increased garbage collection (GC) latency in our search cluster. (This does not affect incoming recording or related data.) To make the application more available in the short term, we have pending internal changes to the application to become more tolerant of GC pauses. We are also provisioning a larger cluster in addition to other technical enhancements to mitigate the root causes of this GC latency. We anticipate shipping these changes by the end of day.
Posted Nov 22, 2016 - 14:13 EST
Investigating
We are experiencing an elevated error rate for searches and session indexing. Users may receive errors while attempting to load session lists (e.g. "Failed to load user list.") and new sessions may take longer than normal to appear.