This issue has been resolved by upgrading the session store's available memory.

It appears that a side effect of the previous issue (https://status.pusher.com/incidents/n10xv45b30bg) was that our dashboard's session store ran out of memory. We are investigating solutions to this.