Last modified 3 years ago Last modified on 01/16/18 15:11:07

dCache Tier I meeting January 16, 2018

[part of a series of meetings]

Present, NDGF(Ulf), PIC(Elena), KIT(Xavier), BNL (Iris)


(see box on the other side)

Site reports


Xavier had nothing specific to report.


Ulf reported that NDGF has been pretty quiet.

They rebooted their head nodes, without any down time. dCache behaved correctly and Nagios reported no problems.

They also took the opportunity to update to the latest dCache version: v3.0.38.


Unfortunately Elena was disconnected shortly after and was unable to reconnect.


Iris reported that of the three dCache instances running at BNL, one is rather old (1.9.x).

They are currently planning how to bring this dCache instance up to a more recent version.

She asked if there's any particular problem with just upgrading dCache.

Two potential issues:

  1. PNFS support is completely removed in all supported dCache versions. This means the namespace must be migrated to Chimera *before* upgrading dCache.
  1. Some versions of BerkeleyDB stores data in slightly different formats. Normally, the BerkeleyDB takes care of this and migrates the database files automatically; however, it isn't guaranteed that the BerkeleyDB version used by the latest dCache pools will be backwards compatible with the BerkeleyDB library used in the old dCache version. This needs to be checked.

Iris will investigate the plan for the namespace conversion (PNFS to Chimera) and whether BerkeleyDB is used to store pool metadata and open a support ticket if there are any problems.

Support tickets for discussion

[Items are added here automagically]


Next week Tigran and Paul are at the XDC Kickoff Meeting in Bologna, so that meeting is cancelled.

The next meeting is at the same time, but in two weeks time: Tuesday 2018-01-30.