Last modified 13 years ago Last modified on 10/19/07 18:40:25

dCache, SRM 2.2 deployment meeting Friday Oct 19, 2007

Participants : Doris, Lionel, Flavia, Timur, Patrick

A first version of the SpaceManager configuration documentation can be found here. With Timurs comments :

I am working on the new version of srm documentation chapters for the 
dCache Book, and the first draft will be available next week. I attached 
the Space Manager chapter, which should explain how the space manager and 
link groups should be configured. I think this new ways to configure link 
groups and VO access is easier to understand than our first version. 
Please not that these instructions apply only to the latest release 
of dCache. Please let me know if you need further help. All comments 
are extremely welcome. 

SRM 2.2 issues

  • Doris : what are the defaults for the link group attributes ? Seems that online=false/other=true. Timur recommends to specify everthing as needed.
  • Doris : will send strange error messages if directory tags are used to overwrite default RP/AL.
  • Doris : Mixing space reserved pools and 'restore' pools. This needs some more thinking.
  • Doris : RP output is not explained because WLCG defined not to use it. It will be used by the Resilient Manager. Timur will add some word on this.
  • Flavia will test against gridKa first to verify that there are no issues left. Lionel and UCSD will follow.
  • Flavia : Ping is not returning the patch level. We will check if we can make this available to the SRM level.
  • Flavia : DESY will provide information again to track the failing requests. (Use case test still failes at FZK and DESY)
  • DESY AND FZK : replica online returns 'no free space', which is understood. FZK and DESY need to change to the new configuration files.
  • Doris agrees to provide the final setup and maybe provide information on all the problems she has been running
  • We need to find a way to make sure that the system repository is consistent after the upgrade.
  • We need to make sure that no files are supposted to go to tape on the day of the upgrade.

Additional issues not directly related to 2.2

Coordinating the current common service challanges.

  • Most Tier I's are still running rather old patch releases except for fzk. Flavia will try to get this coordinated.
  • Strategy on how to make IN2P3 more efficient.
  • Flavia is collecting the precise sequence which causes the atlas problems at IN2P3 and others.

Last modified by patrick @ Wed Mar 3 06:19:07 2021