CHPC Downtime - Tuesday August 6, 2019 starting at 8am

CHPC Downtime  -- Tuesday August 6, 2019 starting at 8am

This downtime is to update CHPC’s core networking switch as well as to add a second Palo Alto firewall in the protected environment (PE) for redundancy.  The work on the core networking switch, estimated to take about 1 hour, will be completed first, followed by the work on the Palo Alto, which is estimated to take about 3 hours.

Along with the networking work, the windows servers, both Beehive in the general and Narwhal in the protected environment, will be updated.  Additional work, described below, on Narwhal will impact the way users access this system.  These systems will be unavailable for most of the day.

Finally, several general environment virtual machines will be relocated from the old to the new VM farm, including the community MySQL and the Jupyter VMs.

Impact: 

Starting at 8am, there will be NO ACCESSto ALL CHPC resources, in both the general and the protected environment, including the clusters, storage, virtual machines, and the windows servers.  Users will not have access to CHPC storage from their local resources during this time. There is a reservation in place to drain all of the clusters (notchpeak, kingspeak, ember, ash, tangent, lonepeak and redwood) of running jobs before the start of the downtime. 

Once the work on this switch is completed access to the general environment resources will be restored in stages, starting with the storage.  Note that it may take some time for the interactive and compute nodes of the clusters to be returned to service – this will be done a cluster at a time – and that the access to Beehive will be returned after its updates have completed.

Access to ALL resources in the protected environment will not be available until the addition of the second Palo Alto is completed.  Again, once work is completed, access to the different resources will be restored in stages.

On Narwhal, along with the updates, additional changes will be made:

• The Narwhal compute server will be reconfigured to become a gateway box for PE access, allowing for direct RDP to this server instead of requiring a hop to this server via the login Narwhal boxes (which will then be retired).
• The current user directories on the C and D drives will be merged to a single user directory on the D drive that will contain the user profile along with additional space for application needs.