#fdio-meeting: FD.io CSIT Project Meeting

Meeting started by mackonst_ at 14:08:38 UTC (full logs).

Meeting summary

    1. Maciek Konstantynowicz (mackonst_, 14:08:48)

  1. Agenda bashing (mackonst_, 14:08:57)
    1. Vratko Polak. (vrpolak, 14:11:08)
    2. Inputs from projects - VPP (mackonst_, 14:11:37)

  2. Inputs from projects - VPP (mackonst_, 14:11:45)
    1. Andrew - vpp v23.10 rc2 today - packages are published (mackonst_, 14:12:06)
    2. Tibor - csit ready for rc2 (mackonst_, 14:12:35)
    3. for now it should be okay to move the vpp meeting to US friendly time instead of 5am pacific time every other call (mackonst_, 14:18:50)

  3. Inputs from projects - TSC (mackonst_, 14:19:12)
    1. Vratko - some discussions re budget with LFN finance, no impact on our project, yet (mackonst_, 14:19:45)

  4. E810 flow "offload" tests (mackonst_, 14:19:50)
    1. Kai - need to revisit the old patch (mackonst_, 14:20:27)
    2. [DO NOT MERGE] dpdk: VPP own DPDK flags | https://gerrit.fd.io/r/c/vpp/+/35743 (mackonst_, 14:20:29)
    3. Maciek Kai - review the latest status per old email thread - and report back here on csit call (mackonst_, 14:21:21)

  5. HW accelerator (mackonst_, 14:21:44)
    1. re SPR QAT2.0 - Kai - no traffic seen in csit tests - in intel local environment got it working after resetting PF instances that in turn resets all associated VFs. (mackonst_, 14:23:43)
    2. re SPR QAT2.0 - Kai - this may not work for csit automated tests, there is work on intel side to have it addressed. (mackonst_, 14:23:53)
    3. 39375: fix crypto device init functions | https://gerrit.fd.io/r/c/csit/+/39375 (mackonst_, 14:24:47)
    4. #info re SPR QAT2.0 - Kai - working on restarting individual VFs, so that automated tests could actually run continuously without breaking, see comments to the patch from Peter and Kai et al (mackonst_, 14:25:53)
    5. re SNR QAT2.0 - Kai - should be corrected to "SNR QAT1.7" (mackonst_, 14:26:46)
    6. re SNR QAT1.7 - Kai - HW instance misreported - after a number of t/shooting sessions, and discussions, involving CSIT (Peter, Kai) and Intel eng (Georgii, et al) conclusion is that this platform needs to be replaced. (mackonst_, 14:28:52)
    7. re SNR QAT1.7 - Kai - John DiGiglio is aware of this. (mackonst_, 14:30:40)
    8. re SNR QAT1.7 - Kai - Maciek Kai to coordinate between CSIT and Intel teams regarding the SNR unit swap. (mackonst_, 14:31:05)
    9. re SPR DSA - Kai - waiting for SuperMicro to find out about public link to BIOS v1.4 that enables required BIOS options to activate and make DSA usable for CSIT VPP tests. (mackonst_, 14:33:27)
    10. re E810 GTPU offload - Kai - status unclear (mackonst_, 14:35:38)
    11. re E810 GTPU offload - Peter - status unclearl (mackonst_, 14:35:57)
    12. re E810 GTPU offload - Maciek Kai - work out the status based on the currently running and broken tests relying on GTPU encap/decap offload on E810 NICs (mackonst_, 14:36:44)
    13. re E810 GTPU offload - Maciek Vratko - check for open jira ticket filed against those broken tests, tests are currently removed from job specs as they were broken for a while, they can be per patch tested in case we need error logs (mackonst_, 14:38:05)

  6. CSIT-2310 (mackonst_, 14:38:57)
    1. rc2 today (mackonst_, 14:39:01)
    2. Tibor - starting now (mackonst_, 14:39:11)
    3. Tibor - as per usual 5 runs iterative to pick up on any anomalies before release and report (mackonst_, 14:39:34)
    4. Maciek - rls in 2 weeks 25-Oct, report publish + 2 weeks after that => 08-Nov (mackonst_, 14:40:23)
    5. Vratko - large scale ipsec policy failing on 3n-alt (mackonst_, 14:41:14)
    6. Vratko - rc1 tests - large scale ipsec policy failing on 3n-alt (mackonst_, 14:41:29)

  7. Physical Infrastructure (mackonst_, 14:44:38)
    1. Maciek Peter - icx-d arrived with 48v dc psu instead of 110v ac psu, being swapped by supermicro. so even though is racked and cabled, it can not be powered up. (mackonst_, 14:45:29)
    2. Maciek Peter - Peter working on activating multiple links in 3na-spr with cx7 NICs to pump traffic and test 1.2 Tbps per socket (6 x 200 GbE) (mackonst_, 14:46:32)

  8. Virtual Infrastructure, CI/CD (mackonst_, 14:46:43)
  9. Environment, Framework (mackonst_, 14:46:57)
  10. Trending and Test Failures (mackonst_, 14:47:10)
    1. Vratko - nothing new (mackonst_, 14:47:57)

  11. Analytics, Presentation (mackonst_, 14:48:15)
    1. Tibor - csit-dash - a new graph added to trending with Gbps (mackonst_, 14:48:35)
    2. Tibor - csit-dash - this is in addition to Mpps (mackonst_, 14:48:57)
    3. Maciek Juraj - wondering if it would be possible to get Arm N2 based server into FD.io CSIT labs. N2 introduces variable length vector instructions that VPP benefits greatly from. We would like LFN FD.io community to have a reference platform in LFN FD.io labs to work with, develop for, optimize for and performance benchmark test on. (mackonst_, 14:53:29)


Meeting ended at 14:53:37 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. mackonst_ (47)
  2. collab-meetbot (3)
  3. vrpolak (1)


Generated by MeetBot 0.1.4.