Difference between revisions of "DCache"
(→Access to the dCache) |
(→Workshops and presentations) |
||
Line 109: | Line 109: | ||
* [https://indico.desy.de/indico/event/19920/ dCache workshop] | * [https://indico.desy.de/indico/event/19920/ dCache workshop] | ||
− | * [https://www.dcache.org/news.shtml This] is | + | * [https://www.dcache.org/news.shtml This] is the best link to recent presentations which is kept up to dat e by the dCache developers. |
Revision as of 14:12, 21 August 2018
dCache is a disk pool management system with a SRM interface (currently v1), jointly developed by DESY and Fermilab. It is one of the currently accepted solutions that UK sites involved in the LCG are deploying to provide an SRM interface to the Grid (the other being Disk Pool Manager). This page intends to provide information for Tier-2 sites who are deploying dCache as their SRM. Installation/configuration/administration instructions will all appear here in due course.
Contents
Installation
- dCache Book is least old page for installation.
- dCache sources and [http://www.dcache.org/manuals/dCacheSoftwareLicence.html license
Upgrading to a new release of dCache
This can be found of dCache page but varies depending on which release you are upgrading from and to which release you wish to upgrade.
Chimera
Troubleshooting
- dCache FAQ
- dCache Problems and Workarounds
- dCache bugs in CERN Savannah
- dCache4SiteAdmins.pdf: A very useful document, giving fairly detailed infor mation about all parts of the dCache system.
- dCache Book: From the dCache developers.
Common administration tasks
- dCache Administration Interface
- Disk Pool Contents
- Renaming a dCache pool node
- Draining a dCache pool
- dCache Administration Scripts
- Splitting the dCache head node
- Upgrade Experience
Database administration
Log files and database
- dCache logfile output level
- dCache Logs
- dCache Log Message Archive
- dCache Billing Information
- dCache Security Service Challenge
Monitoring
SRM v2.2
- dCache SRM v2.2 Status
- dCache SRM v2.2 Testing
- dCache SRM v2.2 GFAL Testing
- dCache SRM v2.2 Configuration - How to setup your SRM2.2 server
VOMS
Generic Information Provider (GIP) plugin
Optimisation
- Optimising dCache Performance - see also Performance and Tuning for general tips and tricks
Access to the dCache
Various different access methods can be used, from FTS service to a CLI such as gfal-* commaands. SOme other useful methods for debugging locally are below:
- Globus Commands
- SRM commands
- dcap commands
- Removing files from dCache
- Latest srmcp client from FNAL can be found here
Documentation for gfal-* commands can be found here and replace the lcg-* coammands for most use cases.
Heirarchical Storage Manager (HSM)
One of the advantages of using dCache to manage a sites storage resources is that it can act as a front end disk cache behind which sits some hierarchical storage system, such as a tape robot (Enstore, HPSS...). While this is not necessarily relevant for Tier-2 sites which do not (typically) have tape, this functionality can be useful in order to optimally utilise available storage resources.
- Using dCache with an HSM.
Useful resources
- GRIDPP-STORAGE archive - mailing list for the discussion of storage issues.
- GridPP storage bug/action tracker.
- GridPP2 Storage - useful information about storage issues in general and dCache in particular, including HowTo.
- eScience Storage Management (limited dCache information).
- LCG - SE_dcache - Generic Configuration Reference
- SRM working group - information on the SRM protocol.
- pnfs - this provides a single namespace used to combine several hundred pool nodes.
- Deployment status of LCG sites.
- Recognise SRM - Whether a remote SRM is a dCache?
Other institutions/organisations experience of using dCache
Workshops and presentations
- dCache workshop
- This is the best link to recent presentations which is kept up to dat e by the dCache developers.
This page is a Key Document, and is the responsibility of Brian Davies. It was last reviewed on 2018-07-25 when it was considered to be 40% complete. It was last judged to be accurate on (never).