IT-protoDUNE coordination (Single Phase and Double Phase)

Thursday 12 Jul 2018, 15:00 → 16:00 Europe/Zurich

31/S-028 (CERN)

Xavier Espinal Curull (CERN)

Coordination: JIRA, Minutes, etc. https://twiki.cern.ch/twiki/bin/viewauth/ProDUNEIT/WebHome

● Round-table: status reports

Round table:

Stu:

- FTS transfers started (also some tests with Condor?). Exploring options and different setup to tune things.

- There might be some issues with WAN connections from Fermilab, but they need to be explored further

Denis:

- Started Configuring the (EOS?) local buffer last week.

Q (Xavi): How much buffer do you need for the Cosmic exercise?

A (Elisabetta): Not detailed yet but we have an aggregate estimation for the needs until April next year of 2.5 Petabytes, which accounts for several contingencies.

Q (Xavi). Will that work for up to a week?

A (Elisabetta): Maybe even more.

Stu:

- Testing CERN/Fermilab network links, and performing parameter tuning. Achieved ~1.2Gbytes/sec with relatively high variance, and packet lots of up 10^-4(%?). The tests were done with xrdcp single threaded running from several nodes/files in parallel

- Next step would be to perform bulk transfers (increasing the number of streams) and revisit the exercises done during the second data challenge

- As at the moment data is "garbage" data, several transfer nodes have been shut down until issues are solved.

Geoff:

- 800TBytes of extra disks have been received (plus a new quad with no disk attached). Some reshuffling will be needed to allocate them

(Stu) remarked to maybe move the XS server to avoid overload.

Q (Xavi to Stu): Would it be useful to grant you access to another node with hyper3?

A: Yes, it would be helpful it it's possible

- Looking for grounding problems, which is requiring to have several systems powered down (around cryogenics).

- Had a cabling issue in one of the switches. It was proposed by the Service Desk to run another fiber, but it's not up to us

- The caverns will start filling up Tomorrow which will cause a restriction of access to the top part of the cryostat.

15:00 → 15:20 Coordination update

Speaker: Xavier Espinal Curull (CERN)

15:20 → 16:00 Round-table: status reports

Speaker: All

Round table:

Stu:

- FTS transfers started (also some tests with Condor?). Exploring options and different setup to tune things.

- There might be some issues with WAN connections from Fermilab, but they need to be explored further

Denis:

- Started Configuring the (EOS?) local buffer last week.

Q (Xavi): How much buffer do you need for the Cosmic exercise?

A (Elisabetta): Not detailed yet but we have an aggregate estimation for the needs until April next year of 2.5 Petabytes, which accounts for several contingencies.

Q (Xavi). Will that work for up to a week?

A (Elisabetta): Maybe even more.

Stu:

- Testing CERN/Fermilab network links, and performing parameter tuning. Achieved ~1.2Gbytes/sec with relatively high variance, and packet lots of up 10^-4(%?). The tests were done with xrdcp single threaded running from several nodes/files in parallel

- Next step would be to perform bulk transfers (increasing the number of streams) and revisit the exercises done during the second data challenge

- As at the moment data is "garbage" data, several transfer nodes have been shut down until issues are solved.

Geoff:

- 800TBytes of extra disks have been received (plus a new quad with no disk attached). Some reshuffling will be needed to allocate them

(Stu) remarked to maybe move the XS server to avoid overload.

Q (Xavi to Stu): Would it be useful to grant you access to another node with hyper3?

A: Yes, it would be helpful it it's possible

- Looking for grounding problems, which is requiring to have several systems powered down (around cryogenics).

- Had a cabling issue in one of the switches. It was proposed by the Service Desk to run another fiber, but it's not up to us

- The caverns will start filling up Tomorrow which will cause a restriction of access to the top part of the cryostat.