Changes between Version 14 and Version 15 of GENIRacksHome/ExogeniRacks/AcceptanceTestStatus/EG-MON-3


Ignore:
Timestamp:
03/15/13 15:58:29 (11 years ago)
Author:
Josh Smift
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • GENIRacksHome/ExogeniRacks/AcceptanceTestStatus/EG-MON-3

    v14 v15  
    2525|| 4          || [[Color(green,Pass)]]          || 2013-03-14           ||               || ||
    2626|| 5          || [[Color(#63B8FF,In Progress)]] ||                      ||               || ||
    27 || 6          || [[Color(green,Pass)]]          ||                      ||               || ||
    28 || 7          || [[Color(yellow,Complete)]]     ||                      ||               || We should re-test this if the switch gains the ability to track MAC addresses ||
    29 || 8          || [[Color(#63B8FF,In Progress)]] ||                      ||               || ||
     27|| 6          || [[Color(green,Pass)]]          || 2013-03-12           ||               || ||
     28|| 7          || [[Color(yellow,Complete)]]     || 2013-03-14           ||               || We should re-test this if the switch gains the ability to track MAC addresses ||
     29|| 8          || [[Color(green,Pass)]]          || 2013-03-15           ||               || ||
    3030
    3131== High-level description from test plan ==
     
    12511251Those are all from the management VLAN, on the port that connects the switch to the control switch; I asked xo-bbn@renci.org if that was expected.
    12521252
    1253 '''2013-03-14 update:''' Chris Heerman confirms on exogeni-design that "you shouldn't see MAC addresses from dataplane interfaces because they're OpenFlow enabled with 'no learning'". (https://mm.renci.org/mailman/private/xo-bbn/2013-March/000041.html, visible only to xo-bbn list members.) We've asked if this is the only way it can be done, or if a different configuration would allow the switch to keep track of (but not act on) which MAC addresses are on which ports; if there turns out to be a way, we should re-run this test.
     1253'''2013-03-14 update:''' Chris Heerman confirms on xo-bbn that "you shouldn't see MAC addresses from dataplane interfaces because they're OpenFlow enabled with 'no learning'". (https://mm.renci.org/mailman/private/xo-bbn/2013-March/000041.html, visible only to xo-bbn list members.) We've asked if this is the only way it can be done, or if a different configuration would allow the switch to keep track of (but not act on) which MAC addresses are on which ports; if there turns out to be a way, we should re-run this test.
    12541254
    12551255== Step 8: verify active dataplane traffic ==
     
    12921292
    12931293I asked xo-bbn@renci.org about this.
     1294
     1295'''2013-03-14 update:''' Chris Heerman confirms on xo-bbn that "show interface port <range> bitrate-usage" is a good way to see usage, and recommends "41-50,64" as the range, to get all the workers' dataplane interfaces plus the uplink towards the GENI core. I tried that, and saw:
     1296
     1297{{{
     1298bbn-8264.bbn.xo>show interface port 41-50,64 bitrate-usage
     1299Utilization statistics for rate. Press CTRL+C to stop:
     1300                   In             Out
     1301Port 41:          41Kbps          14Kbps  /
     1302Port 42:           3Kbps           3Kbps  -
     1303Port 43:          52Kbps          18Kbps  \
     1304Port 44:          24Kbps           4Kbps  |
     1305Port 45:           2Kbps        5403Kbps  /
     1306Port 46:          45Kbps         144Kbps  -
     1307Port 47:        5411Kbps           2Kbps  \
     1308Port 48:          66Kbps          19Kbps  |
     1309Port 49:           0Kbps           0Kbps  /
     1310Port 50:           0Kbps           0Kbps  \
     1311Port 64:           0Kbps           0Kbps  |
     1312}}}
     1313
     1314So that's kind of interesting: There's 5 Mb/sec on port 45 and 47, but those aren't the ports corresponding to my workers. Unless we're just wrong about what's connected to what.
     1315
     1316I tried killing my iperf source, and after a minute or two:
     1317
     1318{{{
     1319bbn-8264.bbn.xo>show interface port 41-50,64 bitrate-usage
     1320Utilization statistics for rate. Press CTRL+C to stop:
     1321                   In             Out
     1322Port 41:          72Kbps          30Kbps  /
     1323Port 42:          38Kbps           5Kbps  -
     1324Port 43:          18Kbps          11Kbps  \
     1325Port 44:          26Kbps           5Kbps  |
     1326Port 45:           2Kbps           0Kbps  /
     1327Port 46:          47Kbps         197Kbps  -
     1328Port 47:          17Kbps           4Kbps  \
     1329Port 48:          14Kbps          10Kbps  /
     1330Port 49:           0Kbps           0Kbps  -
     1331Port 50:           0Kbps           0Kbps  \
     1332Port 64:           0Kbps           0Kbps  |
     1333}}}
     1334
     1335So, that was clearly my traffic. What's it doing on those ports?
     1336
     1337Ah: Physical inspection shows that the hosts are connected to the wrong ports -- the cable from bbn-w6 is in port 45, not 46, and bbn-w8 is in port 47, not 48.
     1338
     1339So, given that, we can see the rate of the traffic flowing between those ports, and I think this step is all set.