News
Important Differences between t1-data and Ceph
With our move to the new Ceph storage, there are some important differences from the old /t1-data
:
- There are no users quotas for the Ceph; all quotas are per-project:
- Project quotas can only be increased by supplying a PO for payment
- If your project runs out of quota then nobody will be able to work until it's increased; make sure there's enough free quota for everyone in the project to add new data and to store your temporary working sets
- You can get a summary of quotas with the
getquota
command. - File paths which previously started
/t1-data
will need to be updated; we recommend using the following symbolic links, which have not changed: /databank
/miseq
/project
- Projects (except
fgenomics
) can only be accessed using/project
- The
fgenomics
project is restricted; if you need data, please contact us via genmail@molbiol.ox.ac.uk - All data is stored in 3 copies at 2 separate physical locations for resilience; there are no separate backups and the old
/t1-data
backups have been switched off - The Ceph has a special feature which removes the need to use the
du
command. To find out how much data a directory contains, use a command in the formls -ldh DIRNAME
, substitutingDIRNAME
for the directory in question. Please be aware that this total may take an hour or two to update when data is added or removed
STORAGE SWITCHOVER TIMELINE
We're in the final stages of commissioning our new Ceph storage, which replaces /t1-data. This web page is a summary of the key dates as we work up to the final switchover.
The Ceph is built using brand new hardware and industry-leading resilient storage software, offering greater performance, easier maintenance, and improvements in reliability.
The switchover also involves an important change in the way quotas work. On /t1-data, quota was per-user and could be increased on request, whereas on the Ceph quota is per-project and can only be increased upon receipt of a PO. This makes it important to have sufficient quota for both the data you already have and the data you'll be creating (including temporary data).
Timeline
January 12th
- Storage switchover and shutdown of "fgenomics" project announced
January 16th
- Final acceptance testing begins
January 17th
- Begin regular copy of newly modified data from /t1-data to the Ceph
January 26th
- All access to "fgenomics" project suspended to allow copying of moved data to Ceph
February 3rd
- Final acceptance testing ends
5pm Friday February 10th
- Cluster shutdown for switchover
Downtime and switchover
- Final sync of project data from /t1-data to Ceph
- All projects move to Ceph, accessible via /project (except "fgenomics")
- All projects removed from /t1-data (except "fgenomics")
- File path /t1-data/project permanently retired
- /t1-data becomes read-only
- Read-only access to /t1-data/project/fgenomics restored
7am Monday February 20th
- Cluster available for use
- Grace period for project quotas begins - additional quota may be added on request and PO raised afterwards
March 6th
- Grace period for new project quotas ends
- Unpaid project quota removed
- All quota must be paid in advance
- Running out of quota will prevent anyone in the project from writing data until a PO is raised