wiki:developers-meeting-20090902
Last modified 11 years ago Last modified on 09/02/09 18:17:02

[part of a series of meetings]

Participants

Agenda

[see box to the side]

Allowing configurable file replicatation for the ATLAS hot file use

The resilient manager could become very attractive to the Altas clouds if it would be able to replicate cached+sticky files in addition to the already supported 'precious' ones. Moreover it would have to make sure that the destination files (replicas) are cached only. Those files (replicas) should be removed if space is running short (which would confuse the replica manager in its current version). There might be other implications, only Alex can tell us. It would be good if FERMIlab could make a statement if they are interrested in such an extention of the RM functionality for Atlas (in particular for BNL) and if yes what the time schedule for such a development effort would be.

Status of work for 1.9.5

A (quick?) review of activity needed for the 1.9.5 release

New info service/info provider

Work is progressing.

Active/passive fixes for SRM client

Dmitry to take care of this.

Moving tape protection inside pool manager

Done.

PnfsManager based listing in SRM

No comment from anyone.

Refactoring of Pin manager

Done.

Terracotta and SRM

Original patch contains about 6,000 lines. Through process of review. Managed to do 2,000 patches.

Waiting on the review process.

Timur submitting new patches

PnfsManager based listing in dirDomain

Done.

PnfsManager based permission handling in all doors

Done.

SRM rm based on ACLs

Timur: SRM-rm was modified by Gerd.

Timur: to investigate.

What happens with PnfsManager without ACLs

..

new http door with https support

Tigran: status of last Friday, able to verify user certificate.

Patrick: Timur, do you have knowledge of groups that would find HTTP beneficial. Timur: once the functionality becomes available I can imagine people will use it.

xrootd mover reimplementation

Tirgan: new mover doesn't work as expected. Gerd is still trying to fix it. First implementation will emulate existing behaviour: one thread per mover. Mover engine does all the IO. Next step will be to redesign the mover part of the pool to remove the thread <--> mover. Current we have 4 threads per mover, now there's a single thread per connection.

the p2p trigger-on-load

Done, committed.

Tigran: can Fermi deployment. End of September for a Fermi.

Issues from yesterday's Tier-1 meeting

No searous issues

Supporting user-triggered replication

For ~1TB of files they would like to have replicas created.

Atlas would like to have

Options are:

  • migration modules
  • hopping on arrival
  • replica manager

Replica manager:

  1. what needs to be changed to support ATLAS use-case
  2. whether there will be time to implement this.

Starting October, Alex is working 10% on dCache and only in a supporting role.

Didn't plan to do any work on resilient manager as not used by many sites.

Timur: can someone write a document? Patrick: Gene has such a document.

Some issues from Karlsruhe, Paul to follow up.

Outstanding RT Tickets

[This is an auto-generated item]

RT 4571: PnfsManager set log slow threshold

No progress; dropped from future discussion.

RT 4699: GGUS-Ticket-ID: #50779 ASSIGNED to dCache Developers srm-{get,set}-permissions does not work on DPM

Put in our functional test suite.

Which version fixes this problem? All branches

RT 4701: feature request: catch java ConnectionException? when SRM-server is not running

No progress; dropped from future discussion.

RT 4702: feature request: pools smaller than 4 GB

No progress; dropped from future discussion.

RT 4706: Default owner for the chimera database has administrative privileges

No progress; dropped from future discussion.

RT 4708: XML parser vulnerability

No progress; dropped from future discussion.

RT 4712: retry setting for suspended files

RT 4716: list 10k on chimera as root takes O(2-4) more than as a regular user!

RT 4721: Problem in GridftpClient? when copying files to dCache with srmcp

RT 4733: gsidcap failures with new CAs

RT 5039: SpaceManager in 1.9.3-p3 not picking up policy directives

Need to drop these parameters from SpaceManagers?.

Going to drop these parameters from

Drop ticket from future discussion.

RT 5055: thousands of mounted pnfs .(access) entries

RT 5067: wrong configuration parameter names for default AL/RP

Same problem as RT#5039.

Drop ticket from future discussion.

Review of RB requests

DTNM

Proposed: same time, next week.