As noted on this blog, we had planned to deploy 1.17 earlier today (February 8, 07:00 UTC). The deployment preparation took us much longer than we anticipated, and once we did attempt to deploy (at around 13:00 UTC), we encountered some unanticipated performance issues. We reverted to the previous version of the software (1.16) and now we’re investigating the performance problems. We do not yet have a schedule for when we’ll attempt another deployment, but when we do, we’ll post more information here.
Update, 16:27 UTC: We have ironed out the performance issue and have resolved a few other immediate problems that arose as well. We are attempting a second deployment now.
Update, 17:55 UTC: Other issues have come up, and we have canceled our second attempt. We’ll investigate further and share our findings. We don’t plan to try to upgrade again today.