wiki:developers-meeting-20101027
Last modified 10 years ago Last modified on 10/27/10 18:01:30

[part of a series of meetings]

Participants

Thomas; Dmitry; Christian, Antje, Jan, Paul, Owen, Karsten, Tanja, Tigran

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 :)

Dmitry: not working on dCache .. working on enstore (commissioning of drives). Christian: EMT, bug tracking, documentation fixing, SSHv2 implementation Antje: Book, Tickets, stuff Jan: knowledge transfer; refactoring; some Book. Paul: Janitor work, SRM-ssl, reviewing. Owen: EMI, test-bed, debugging, CERN-interaction, working with dcap release, tickets; ETICS Karsten: receiving knowledge transfer from Jan; working towards Argus. Tanja: Reading a lot of email, tickets, bothering people with tickets, NFS work Tigran: CHEP; deleting emails. Thomas: CHEP; work on Book; research into GSI with Xrootd.

Plans for patch-releases

Should we make a new patch release?

Just released 1.9.5-23. Need to update the webpage.

wait for Gerard to confirm xrootd bug

Plan to release 1.9.10-2 and make it green at that point.

From CHEP we learnt that most of the sites do not upgrade if the release isn't green. Most don't even test releases if the release isn't green.

Thomas reported that some sites complained about bugs / missing features that have been added in releases after 1.9.5. Sites were under the impression that they are "supposed" to run 1.9.5.

Intermediate feature releases make sense for external communities, if not for WLCG communities.

Trunk activity

Progress with new features...

SRM is broken; the auth-persistence-manager bean cannot be instantiated. This is the org.dcache.auth.persistence.AuthRecordPersistenceManager?. Constructor fails due to reflection failing.

Review request

Owen asked for people to review dCacheConfigure. This is important for automatic testing of dCache.

Issues from yesterday's Tier-1 meeting

Replica Manager

Antje: working on the installation chapter of the book. In that chapter, it mentions that the replica manager requires certain database configuration. Since this isn't needed I wanted to move it to the replica manager chapter. Looking at this chapter, it's old (out-of-date).

Email to team asking who could update the replica manager chapter. Tigran + Antje will update this chapter.

Are there any instances? Yes, three: Fermilab, BNL and DESY. DESY instance will die to two months from now but Fermi, at least, will continue using it.

Outstanding RT Tickets

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

RT 5856: fqan substitute in LinkGroupAuthorization?.conf

Dmitry will have a look into this.

RT 5859: xrootd not working when multiple streams are requested neither with user code

Merged into 1.9.5, 1.9.7, 1.9.8, 1.9.9. Latest 1.9.5 has been released as 1.9.5-23. Waiting on Gerard to confirm problem is fixed.

RT 5876: Srm Client throw Exception without any hint what went wrong

Dmitry will have a look.

RT 5882: moving head/single/pool.conf to some example directory

No progress. Don't need to discuss any more.

RT 5886: extra info in paths.properties

Owen to talk to Tigran about this offline. No need to discuss further.

RT 5900: Problems with SRM 3rd party transfer

Waiting on user input. No need to discuss further.

RT 5919: Utility domain in 1.9.10-1 writing into /opt/d-cache

Two of the files (createDDL.jdbc and dropDDL.jdbc) are created due to a bug introduced by Thomas.

Location is controlled by the file persistence.xml. The problem is we can't get an arbitrary value. We can hard-code this to be /tmp

Dmitry has looked for an option to switch off creating the table, but hasn't found a switch that works. The switch is in our persistency.xml file, but doesn't seem to work.

Can we replace toplink with jpox?

History: Timur tried to use jpox and it didn't work and switched to toplink. The situation may have changed for jpox, so it might be worth revisit this issue.

The move from toplink to jpox would fixed then the SRM would work on OpenJDK.

At least jpox doesn't try to create the tables that already exist, unlike toplink.

Also, if I want to add index to column, make an annotation (or in persistence.xml file) to effect the change.

Can you give JPOX the javax.sql.DataSource? object instead of the JDBC string? Don't know.

Review of RB requests

DTNM

Proposed: same time, next week.