wiki:developers-meeting-20101110
Last modified 10 years ago Last modified on 11/10/10 18:05:10

[part of a series of meetings]

Participants

Thomas; Gerd; Dmitry, Patrick; Antja, Paul, Owen, Tanja, Christian.

Agenda

[see box on the right-hand side]

Postcards

Thomas: still working GSI support for xroot. Mostly finished. Up for review soon. Dmitry: Nothing interesting .. a couple of interviews. Patrick: HEPix, visiting Alex Sim and Arie ? talking about future of SRM; planning the Prague meeting; contact with Jens about 1.2.1 deliverable. Gerd: Most of the time being sick. Upgraded all the pools, small bug in p2p ping msg, second time due to a FTP bug and started to be a problem. Quite some time with Alice tape management .. not much developing: mostly gPlazma. Antje: working on the book. Testing fresh installed 1.9.10 after the release; passes G2 tests, haven't done the S2 tests yet. Paul: EMI work (SRMSEC), reviewing, tickets, helping. Owen: found parameters to make EMI-ETICS build all our components, started gLite release of dcap. Tanja: Tickets, working of NFS. Christian: working on SSH code (a little) key-based auth. almost working. Lots of EMI-ETICS stuff (mapping things). Working on the documentation. Bug-state mapping work. Karsten: work on Argus

Plans for patch-releases

Should we make a new patch release?

Trunk activity

Progress with new features...

Issues from yesterday's Tier-1 meeting

Gerard webdav

username + password

Some of the problems we've been having may go away with updated Milton library.

TODO:

Patrick talk to Gerard about webdav.

In Sweden, some of the money they got for the national storage facility are to update browsers to support certificates.

We can try to integrate the new Milton library before Dima does his testing.

Can we provide a reference end-point for people to test clients against?

Issues from EMI

Bug tracking states

Bug status mapping between our bug-tracker and EMI's concepts.

It's a complicated issue, so discussed off-line.

ETICS

We now have 1.9.5 compiling.

There does seem to be a problem trying to upgrade libraries within ETICS.

SRM client and libdcap

There are some tickets about the SRM client (memory client)

We have a fix for this, but committed only to trunk. Trunk has not been released.

Who will release srmclient?

Gerd: there have been several fixes to the SRM client that are only in

Action:

Paul: talk to Tigran to ask about back-porting fixes to 1.9.5 branch and releasing from 1.9.5

Who can release srm-client and dcap library when Owen has left?

Tigran to answer this when Owen has left.

When do we release srm-client? We can try to fit all outstanding the patches into the next release.

Probably ~20 commits to be back-ported. These are only patches that touch ONLY the srm-client module.

Tigran: the blocking factor for releasing a new srm-client is the testing the SRM client.

Some people working on "Site Acceptability Monitor" are using our client.

Dmitry has a script for testing SRM .. he'll send this to Owen to be included in the Hudson test process.

Dmitry to collect a list of patches that we should back-port to 1.9.5.

SRM recursive listing

A number of patches to extend the list handler to do

Pipelining in SRM lookups.

Doing recursive inside the list handler allows pipelining to span different directories.

SRM spec returns a detail field

The easy solution is to return the directory simply as a long list.

DPM doesn't support "recursive ls".

Dmitry suggests discussing this with SRM developers.

Patrick sends Gerd the address to send the email. Gerd to send the proposal.

HEPix

"" Most of the time, people gave site reports. Mostly on hardware pledges.

NFSv4.1 Because those people are sys-admins, they are keen. They was a lot of interest.

Presentation on EOS. This presentation has been listing all the dCache features. "I can imagine you did research to find out which storage-element already provides these features.."

It seems nobody has done this survey. Ian Bird just ignored query why they are re

Claim dCache cannot do 2,000 requests per second.

Cornell moved away from dCache to xrootd. Unfortuantely, they lost the onbly dcCahe person. The easiest thing was to move to xroot as they already have two people who understand it.

Have been talking with the Fermi ScientificLinux? people. They were very helpful. They will help in getting Tigran's patch into SL. It will be a separate RPM, but will be part of the SL release.

Troy: whenever they do something, Redhat looks into this too. This may trigger RH to take Tigran's work.

Future of SRM

Because LBML is not getting any more money for SRM they can't do any more work on it.

Arie is preparing for a new call. Guess this will be on how applications can get access to data.

His plans to make the SRM useful without changing too much.

Plan is to write a use-manual.

The BestMan? support will continue. Continue the specification isn't supported.

OGF is the group managing the SRM. Jens is the only person left in that group.

The majority of the work is "us": the four main storage elements.

The specification is not clean ("intrinsically broken").

To make it useful, WLCG made agreements.

Some functions that aren't used (e.g., move all my data from T0-->T1; ...)

Jens: all the agreements have been fed back to the OGF web-page. You can also find them on the LBLL SRM web-pages.

Outstanding RT Tickets

[This is an auto-generated item. Don't add items here directly]

RT 5768: java.lang.IllegalArgumentException?: Name Mismatch in 1.9.9-1 pool

Reassign ticket to Gerd.

Also need the other files ending ".pre-197"

Review of RB requests

DTNM

Same time, next week.