[part of a series of meetings]

Participants

Christian, Jan, Paul, Tigran, Karsten, Tanja.

Agenda

[see box on the right-hand side]

Postcards

Up to two minutes (uninterrupted) per person where they can answer two questions:

• What I did last week (since the last meeting),
• What I plan to do in the next week.

No questions until we get through everyone :)

Jan: nothing but pampers.

Christian: pampering EMI, working on our test suites (G2), partially successful in getting G2 working; trying to run these tests against EMI-1-RC4; problems caused by bdii. EMI release will be "sometime next week" (honest govn). Wanted us to deploy Amga here at DESY + other stuff but EMI has melted my brain.

Tanja: fixing myself .. results are promising.

Karsten: new patch changing how gPlazma-2 plugins receive their configuration, now on RB. Worked on the WLCG poster.

Tigran: write script for DESY people to dump Chimera. They want to take two sub-trees from two different, existing Chimera trees and create a third Chimera instance. Yesterday and today NFS code is merged into dCache (trunk + 1.9.12). The motivation is to fix compatibility with latest Linux kernels (>= 2.6.39). Added to NFS code support for an external mapping service (ie. talking to gPlazma). Prepared presentation on <<a random walk on map-reduce>> for physics-computing. Karsten to present Argus. Updated RB to new version .. expect problems ;-)

Paul: looking at fix a bug reported about dccp with dcap++ enabled (no, this doesn't make sense). Fix is easy, proving it's fixed is a little harder. Also looking at SRM and overwriteMode property. SARA have a GGUS ticket about FTS no overwriting a file in dCache. This is because FTS uses the overwriteMode flag to say "overwrite this file". dCache behaviour at SARA is to ignore this flag. Patrick and me are investigating whether enabling support for overwriteMode will work.

Plans for patch-releases

Should we make a new patch release?

Nope.

Trunk activity

Progress with new features...

gPlazma2

Karsten .. changing the interface of the plugins to use Properties. This is a prelude to allowing better integration with our configuration system.

Tigran found a problem with the Argus plugin.

Currently we cannot verify banning people with Argus.

Would be nice to have a machine for installing external dependencies for internal testing.

Do we still have any use of Austrian? It's currently offline; so we can remove it from Jenkins (nee Hudson).

Jenkins / Hudson.. they're still binary compatible. Jenkins is the more active of the two. There are some new plugins, such as ones for bringing up a VM and run a test on that VM image.

A couple of days ago; Tigran complained that gPlazma-2 treats everything as users, Gerd patched this, but the it's still a little ugly and we're waiting for the API to stablise.

gPlazma and VOMS

How about a plugin that asks VOMS server(s) if the user comes in with a X509 certificate but no FQANs. This was a request from someone.

One way around this is GridSite? who have a plugin for Firefox that will generate a proxy-certificate and upload it if the website identifies that it wants this. Our WebDAV pages don't advertise for this and wouldn't understand the upload anyway.

We can talk about this next week, during the dev. meeting.

SVN commit messages

We have a commit-log templates, but not everyone is using this. How do we enforce this? This is important as not doing this creates a lot of work for Christian.

Tigran wrote a tool for automatically generating these messages.

Paul wrote a parser for log-messages to verify format. This could reject commits if they are badly formatted.

Could we do something for next week's dev. meeting; e.g., a short talk and demo of Tigran's tool?

Christian to email Dmitry with suggestion.

How to handle documentation updates

We decided previously that no changes go into dCache code without either updating documentation or opening a ticket (describing that documentation needs to be updated) to make sure this happens. This currently doesn't happen. This is another topic we could discuss during the developer meeting next week.

IntenseDebate is proving quite useful; there have been a fair number of problems identified that Antje has fixed.

dCache dev. meeting next week

Reminder: it's a little late, but please email Dmitry if you have any ideas for items.

Issues from yesterday's Tier-1 meeting

GridKa? and Triumf

Issues from EMI

Nothing that regards us. We've done the FHS-compliant build; it's in EMI-1 RC4 (RC3 is closed) as dcache-server-1.9.12-2 .. This is built from the 1.9.12-2 tag with the "server-fhs.rpm" ant target.

All configurations in ETICS are now locked for EMI-1 and there's nothing blocking from us on the EMI-1 release.

We are currently "certified" but not yet "verified". This transition requires a manual process from someone who checks our certification paperwork.

The security group has a new version of their SAML-assertion XACML document.

StAR (storage accounting record format) is progressing and getting some feedback from sites.

Outstanding RT Tickets

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

 RT 6167: bringonline failed due to pinning failed:finding read pool failed

It seems the issue is whether to back-port the update to the Pin-manager--PoolManager message; but, without Gerd, we don't know the exact issue here.

Review of RB requests

Done.

DTNM

Next week is all developers in DESY week; so same time in a fortnight.