The SSH queue has remained stable through everybody's top-of-the-hour cron jobs, so we're marking this as "resolved".
Posted over 2 years ago. Jun 12, 2015 - 11:17 UTC
We've experienced a bit of trouble with SSH queueing on one of our load balancers; new connection requests had been timing out or left half-open. The affected process has been restarted, and we're keeping an eye on the situation now. Web access, API access, Git access over HTTPS, Mercurial access over HTTPS, and downloads remain unaffected.