wiki:developers-meeting-20101109
Last modified 10 years ago Last modified on 11/09/10 16:45:52

dCache Tier I meeting November 9, 2010

[part of a series of meetings]

Present

dCache.org(Tgran,Tanja,Paul), IN2P3(Yvan), Sara(), Triumf(Simon), BNL(), NDGF(), PIC(Gerard), GridKa(Doris), Fermi(), CERN(Andrea)

Agenda

(see box on the other side)

Site reports

CERN

Everything going well.

KIT

Everything is fine.

PIC

Everything is OK at PIC.

Security

How to set username+ password authentication?

As Gerd for help with documentation.

VOMS for authenticating users.

Talk to Dima about getting information about clients.

IN2P3

Thumper machine: no news from

Still have problem with ATLAS export pools. 7 pools that are current

Most activity is on two pools.

Tired to show on some plots number of active connections on these pools.

400 connections that are active. This is a mixture between GridFTP connections and dcap.

Not checked the IOWAIT activity on the pools; the disk seems to be fine. Have not run IOZone tests.

Have not changed the configuration of FTS yet.

iperf tests between PIC to IN2P3 the results in the tickets. Transfers from Linux machine to another Linux machine was fast. Results for the Solaris machines current waiting.

Run iperf in the port range 20,000 to 25,000; saves firewall issues.

Sys-admins are in contact with Solaris support. Unclear if they will give some suggestions what to change.

Before end-of-November plan to deploy new machines. I plan to use these machines to test ...

Downgrade on one machine; this did not help.

May be a problem with the pool manager which is responsible for choosing which pool to do the transfers. Only a few transfer pools are selected from those transfers.

Pools have the same configuration in pool manager? It is the same pools

Separate issue

MTU change; RFC "MTU path discovery". Requires properly configured routers. RFC ?

Already in SL-5.

Triumf

Last week in production things are OK. Small amounts of GridFTP transfers due to time-outs. Some lcg-cr commands are failing, still failing.

Support tickets for discussion

[Items are added here automagically]

RT 5906: Atlas transfer problem

DTNM

Same time, next week.