Personal tools
You are here: Home / News / Cluster relocation
Log in


Forgot your password?
 

Cluster relocation

Upcoming relocation of the compute cluster and all servers starting june 15th 2012.

Attention

As already communicated through other channels our cluster and all accompanying servers move to a new server room on our cluster.

The relocation will start on June 15th, thus from this point in time we have:

  • Downtime of both login-nodes and all compute-nodes which means abortion of all running computational tasks in the grid-engine and all interactive sessions.
  • Downtime of all Fileservers i.e. BIO2, /data/deep_seq*, /data/<your-group> and /home inaccessible which also affects the Sequencing-Facilities (Chen Lab), usually writing directly to the fileservers
  • Downtime of the BBC-Server including the BBC-Wiki "Plone", the BIMSB-Calendar, SVN-Repository, GBrowse2
The downtime will last until at least June 25th, in case of unforeseen issues, we plan to revive at least some core-parts of the cluster by the 29th of June at the latest.

Even when you do not do computations directly on the BIMSB-HPC-Cluster, the other aspects of the upcoming outage in June will likely affect your work:

  • The Sequencing-Services (Chen Lab) will need to reduce their throughput. (See Mai from Claudia Langnick, May 10th 11:40:31)
  • The BIMSB-calendar will be unavailable.

Please plan your work and computations in a way that all is finished (or can be halted) before June 15th, and have the small fraction of data that you need during the downtime as a copy on your local machine.

What will be still available during the Downtime:

  • WindCompS (the 8 Nodes of the Windows-Compute-Servers)
  •  WCS-Data (only 3TB storage space in total - 900GB still available)
  •  BIO-Folder (for the Rajewsky Lab) 

 

As all data on the servers to be moved need to be backed up in advance, we have a important request for you: PLEASE DO NOT MOVE big amounts of data (greater 100GB) within any of the affected directories (home, group, ..) to a different location; also do not RENAME folders containing this large amount of data. The reason for this is: our backup would re-treat all this data as new data (cannot recognize renamed/moved data), which could take more time than planned for the final, incremental backup and would break all relying time-schedules. Our other option would be to risk to loose your (and all other's!) recently created/moved/renamed datafiles during the relocation - we doubt that YOU would like to be identified to be the reason for this... If you ... - have to rename/move big amounts of data, - need to access some of your files during the downtime, or - have any other question regarding the details of the downtime, ... please to not hesitate to contact us - we'll find a solution for all issues that we KNOW about in ADVANCE. But please also understand that during the downtime we will be extremely busy and might only have little time to find solutions for issues that we get known about at short notice.
Document Actions