UPDATE 9/9 - 5pm - Since the hurricane threat to Tallahassee has continued to increase over the past 24 hours, we are obligated to turn off all RCC services. Please stay tuned to our Twitter feed for updates.
UPDATE 9/9 - 12:30pm - Hurricane Irma continues to pose a greater threat to Tallahassee. Per our earlier message, we will be preemptively turning off the following systems to prevent potential damage:
- Older HPC nodes in Dirac (see 9/8 update for list)
- Spear servers
- We are unmounting Lustre from HPC nodes (but the system currently remains available via Globus).
All other systems, including Panasas, most HPC nodes, and Virtual Machines remain online.
In the event we need to shut down more systems, we will annouce it here, via email, and on and our Twitter feed. Expect an update no later than tomorrow (Sunday) afternoon.
As a precautionary measure, we are turning off some of the older compute nodes in the Dirac Data Center. A list of affected nodes and partitions is at the bottom of this update. We will begin draining jobs from these nodes today, and shut off the nodes no later than Sunday evening, September 10. Any jobs that are still running on these nodes at that time will be killed.
The reason that we are shutting down some of the older nodes in Dirac is to mitigate a potential heat risk in the machine room that caused major damage last year during Hurricane Hermine.
We are also unmounting Lustre from HPC login and compute nodes, effective today. Lustre will remain online and available via Spear and Globus. If we do need to quickly shut Lustre down, this precautionary measure will help ensure that no data loss or corruption occurs.
We have rescheduled our "Introduction to HPC Workshop" for Tuesday, September 19 at 1:30pm.
Our current plan is to keep most of the HPC and other RCC services operational. Services that we plan to keep ONLINE include:
- Most owner-based HPC compute nodes and all compute nodes in the Sliger Data Center
- Panasas filesystem
- Spear Interactive Cluster
- Lustre (not available on HPC, but available via Globus/Spear)
- Virtual Machines ("SKY")
- Research Archival Storage
We will re-evaluate the situation on Sunday afternoon and will post another update. In the event that the storm threat increases in the Tallahassee area, we may elect to shut down additional systems (most likely Lustre and Spear, but possibly more).
If we need to turn off additional systems, or if there is any further developments, change of plans, or problems that arise, we will do our best to communicate as quickly as possible. Keep an eye on our website and our Twitter feed.
Please direct any questions or comments to email@example.com. We will monitor this account over the weekend and early next week.
UPDATE 9/7 10pm - FSU is closed tomorrow and Monday. RCC staff will be on-site tomorrow morning, and will post an announcement in the morning regarding our action plan for the storm.
As Hurricane Irma approaches the Gulf of Mexico, RCC staff are preparing for any possible impact scenarios, including a potential landfall near Tallahassee around Tuesday September 12.
- On Friday, September 8, we will decide if we need to shutdown our older computing infrastructure in the Dirac server room during the weekend; this will mainly impact the backfill and general access partitions.
- On Sunday or Monday 9/11, we will decide if we need to shutdown all services (lustre, some sky, spear) in the Dirac server room.
- We will try to keep our services in the Sliger room up as much as possible, but if warranted, we will shutdown our equipment in that room too.
The workshop on September 12 may be postponed depending on weather conditions. Follow us on Twitter (@fsurcc) for the latest news.