At 17:20 PST our monitoring system reported degraded network performance on nodes v2102, v2103, v2104.
At 17:43 PST our engineers have designed a fix for this issue.
At 18:16 PST we began to gracefully shut down all VMs to apply the fix.
At 18:36 PST fix applied. All VMs will start automatically in the next 15 minutes.