Planned storage array upgrade: storage/praha1/, storage/vestec1/

As part of modernizing our infrastructure, we are migrating to new, higher-performance, and larger disk storage, which will replace the current storage array /storage/praha1 and its equivalent /storage/vestec1 (these represent the same storage array linked via a symbolic link).

The entire migration will proceed in the following steps:

1. Preparation Phase (Oct 1 – Oct 5)

Starting on Thursday, October 1, we will begin the preliminary data synchronization to the new storage array. We kindly ask you to limit the creation of large volumes of new data and extensive file modifications on this storage array between October 1 and October 5. This will significantly help us shorten the time required for the final cutover.

2. Main Outage and Cutover (Monday, Oct 5)

Complete storage unavailability: On Monday, October 5, the storage cutover itself will take place. Please anticipate a multi-hour downtime of the /storage/praha1 (/storage/vestec1) storage.
Compute node outage: We will pause job scheduling on compute nodes located in Prague that use this storage array as their home directory (/home). Other compute nodes in MetaCentrum will continue to run without interruption.
Attention to data from running jobs: If you have jobs running on October 5 on nodes that are not affected by the outage, and these jobs attempt to write or copy output data to the unavailable storage array, the write operation will fail and the data may remain stored locally on the compute node (in the scratch directory). In such cases, you will need to copy the data manually after the outage concludes.
Once we verify the proper functionality of the new mounts, user quotas, and automatic snapshots, the affected nodes will be released back into production.

3. Regular Operation and Final Re-synchronization (approx. Oct 5 – Oct 10)

Following the launch on October 5, you will be able to work on the new storage array as usual. In the background over the next ~5 days, a re-synchronization process will copy any remaining changes made on the old storage array between October 1 and October 5.

Safety of your data during the transfer from the old to the new array:

In the event of a version conflict, your newer files will always take precedence and will not be overwritten by older versions.
Files that you deleted in the meantime will not be deleted from the old array (no data will be inadvertently removed).
The original storage array will remain archived for a period of time to resolve any non-standard issues if needed.

If you have any questions or encounter any issues, please contact user support.

MetaCentrum Team

Ivana Křenková, Tue Sep 01 21:40:00 CEST 2026