Tier-1 reports, 06-06-2006
* upgraded to gLite 3.0
* still need to install gLiteCE
* FTS upgrade complete
* channels to T2 configured
* relocation SC castor to new data center
* Network upgrade to new provider
* LAN refactoring complete
T1 Status:
* 7 CT,
* start from 2006-06-02 18:11:25 to 2006-06-03 11:05:03
* arise from configuration prob, the autoupdate script get stuck in java config (due to the conficts of j2sdk version installed), and have been fixed by rerun the glite yaim config script, for all WNs that split into 12 group.
* 6 JS
* arise from batch server name is different from CE name, that couldnt find available resourece from CE, have been fixed by updating the torque client config file and server name forcely.
* 1 JL
* arise from mixing old gip configurations and new gris config/ldif file, have been fixed by rerun the config_gip function
Lyon, IN2P3
- IN2P3-CC (T1):
all nodes (except SEs) has been updated to gLite3.0. WNs are still in LCG2.7.0, but will be updated this week. Additional lcg CE will also be added this week to prevent the site from CE overloading due to massive job submission.
gLite CE set-up in progress. It seems that there is a wrong merge between M/W environment and local environment (typically with environment variables such as PATH, PERLIB, ...).
FZK
Site Report Tier 1 GridKa (FZK):
MW environment:
- file server problems on Friday, thus batch system was stopped - should work again now.
- gLite 3.0 environment (RB, PX, BDII, SE-Classic, LFC, MON, about 30 WNs, CE, UI) still needs to be tested yet, test phase with users hopefully to be started soon.
- test phase is planned for one more week, gLite 3.0 environment is planned to be moved into production afterwards
- main site admin not permanently available due to vacation next two weeks, delay of gLite 3.0 in production at GridKa seems likely
Service Challenge 4:
FZK started tape transfers for background data flow, currently with 3 dcache writer nodes and 3 tape drives (LTO3). Rates are oscillating between 40 and 60 MB/s. Since Sunday 28/5 peaks above 60 MB/s (1 hour gridview average) are observed. We are investigating the cause for the variations. Last weekend there were no transfers due to problems with SRM. Today 6/6 the transfers started again. The coming week additional nodes and drives will be out into service.
CNAF
INFN-CNAF/Tier-1 report uploaded as ROC:
SC4 preparation, INFN CNAF REPORT:
- optimization of Castor2 configuration still ongoing, as the daily average the last week
is still below the target (200 MB/s)
- FTS production server installation: still ongoing
Sara
Last week we have separated the mass storage environment for
Tier1 and our supercomputers. This operation was
successful.
We have experienced some problems with the FTS transfers agents. These had to
do with
parsing errors of the services.xml file. There was
also a problem with the information system on the FTS server. In version 1.4 we
had to write our own provider but now one is provided by yiam.
The old one was still in place and messed things up.
There is also an issue with our local LFC and VOMS proxies, which we are
looking into now.
RAL
GridFTP servers were running slow
causing timeouts.
Restarts fixed the problem. Currently it has been running for two days with out
problem.
FTS server will be upgraded 11:00 GMT on Monday 5th June.
Some progress on gLite CE but slow due to holidays of 3 key people.