wiki:GENIRacksHome/InstageniRacks/AcceptanceTestStatus/OpenTopics

Version 16 (modified by lnevers@bbn.com, 7 years ago) (diff)

--

InstaGENI Acceptance Tests Open Topic

Open Topics: 2013-02-27

Experimenter Tests

IG-EXP-7: Click Router Experiment Acceptance Test ---> Expected Date: March 01
Niky is modifying click router setup to run in two racks scenario. On 02/26 she was implementing click router environment workarounds. Expect to be able to run test by Friday March 1.

Administrative Tests

IG-ADM-1: Rack Receipt and Inventory Test ---> Expected Date: March 13
Only remaining blocker is government property tracking (specifically tags at BBN).

IG-ADM-3: Full Rack Reboot Test ---> Expected Date: March 11
Need documentation by March 8th, would like to execute on March 11

IG-ADM-4: Emergency Stop Test ---> Expected Date: March 4 (March 5th alternate)
An Emergency Stop Procedure detailed test outline is being defined. The preparation steps are currently being executed and coordination is taking place with ES teams. Provided feedback on the InstaGENI documents, which have been addressed. Trying to schedule IG-ADM-4 with the InstaGENI team for March 4th, with the 5th as an alternate or a re-test date. These dates should work for Eldar and the GMOC. There are no blocking issues at this time.

IG-ADM-5: Software Update Test ---> Expected Date: March 5 (March 11 alternate)

Nick to complete tests the current GENI-recommended versions of FV and FOAM by Friday March 1. Once Nick completes testing and InstaGENI Team requests site admins to upgrade, expect to run upgrade tests on March 4th. If we miss that date, March 11th is the last chance before GEC.

IG-ADM-6: Control Network Disconnection Test ---> Expected Date: March 7 or 8
Expect to execute full procedure on March 7th or 8th.

IG-ADM-7: Documentation Review Test ---> Expected Date: ???

Prioritized outstanding documentation email on 01/31/2013. Waiting on responses from InstaGENI team.

Monitoring Tests

IG-MON-1: Control Network Software and VLAN Inspection Test ---> Expected Date: ???

  • Blocking question in email discussion about source code availability on racks.
  • GPO is blocking and needs to capture questions about control network setup, which will be collected as part of IG-MON-2.

IG-MON-2: GENI Software Configuration Inspection Test Expected Date: ???
GPO is blocking and needs to capture questions about control network setup.

IG-MON-3: GENI Active Experiment Inspection Test ---> Expected Date: ???

IG-MON-5: GMOC Data Collection Test ---> Expected Date: ???

Working on a consolidated debian package of tango-monitor-instageni to be run on control nodes. Package is to be initially tested on BBN head node. Software to be delivered will include:

  • Code to run on one or more head nodes (and configuration of the credentials which that code needs to speak AM API and ProtoGENI data collection API)
  • Update of the tango-monitor-pgenishared package which runs on shared workers.

InstaGENI Site Confirmation Tests Open Topic

Performance Issues:

  • Leigh investigating PC to VM results which did not meet his expectations for bandwidth shaping. In a scenario where no bandwidth is specified in RSpec, nodes will get a 1 GB interface whose traffic is shaped to 100 Mb/s for a VM, but not for a PC. But performance recorded from PC show unexpected traffic levels:

End Points One Client Five Clients Ten Clients
PC to VM 101 Mbits/sec 101 Mbits/sec 101 Mbits/sec
VM to PC 101 Mbits/sec 101 Mbits/sec 101 Mbits/sec
  • Potential Site to Site VM over OpenFlow performance issue for Utah sites. Sent email to Rob. but not sure if he has access to page yet. Performance results captured show the following iperf TCP throughput:
End points One Client 5 Clients 10 Clients(tcp)
GPO IG VM to Utah IG VM 11.2 Mbits/sec 21.7 Mbits/sec 41.5 Mbits/sec
GPO IG VM to Utah PG VM 10.7 Mbits/sec 25.2 Mbits/sec 36.7 Mbits/sec

According to Rob, the only limitation is between the Utah campus switch on campus and the switch in the SALT I2 POP, which are limited to 4Gbps between them. The results we go were substantially lower.

Several fixes have been put in place since the measurements were collected in mid-February. Need to re-ran to experiment to verify that results are unchanged.