#fdio-meeting: FD.io CSIT project meeting

Meeting started by mackonstan at 14:01:38 UTC (full logs).

Meeting summary

    1. Peter Mikus (pmikus, 14:02:10)
    2. Fan Zhang (fan_zhang, 14:02:31)
    3. Juraj Linkeš (jlinkes, 14:02:39)
    4. Tibor Frank (tifrank, 14:02:57)

  1. Agenda bashing (mackonstan, 14:04:15)
    1. Vratko Polak. (vrpolak, 14:04:32)

  2. Inputs from projects - VPP (mackonstan, 14:04:44)
  3. Inputs from projects - VSAP (mackonstan, 14:05:23)
  4. Inputs from projects - TSC (mackonstan, 14:05:34)
    1. Peter as a new committer for ci-managment (mackonstan, 14:05:47)
    2. A new fdio wide mailer created for committers to coordinate election process for FD.io TSC (mackonstan, 14:06:45)

  5. New tests (mackonstan, 14:07:22)
    1. func: flow "offload" patch - Peter, support for multiple nics needs to be added to vpp_device (FVL NICs only today), to support CVL. (mackonstan, 14:10:02)
    2. func: Vratko, Peter - should follow approach used for perf tests, use suite generator to cover multiple NICs, and job_specs for which tests to run. (mackonstan, 14:18:10)
    3. perf: flow "offload" patch - Fan, work in progress (mackonstan, 14:20:51)
    4. IPsec SPD flow cache patches, Juraj, 1 patch merged, 2 other are waiting for VPP patches to merge first. (mackonstan, 14:21:18)

  6. Releases - CSIT-2110 (mackonstan, 14:21:50)
    1. https://wiki.fd.io/view/CSIT/csit2110_plan (mackonstan, 14:22:10)

  7. Releases CSIT-2106 (mackonstan, 14:22:53)
    1. Tibor, re-run iterative tests after the latency fix, results in report master (mackonstan, 14:23:30)
    2. Tibor, also finished iterative af_xdp tests and all remaining coverage tests, results in report master (mackonstan, 14:24:10)
    3. Tibor, All pls review the latest measurement data in latest master, especially latency and af_xdp, so that report can be pushed into CSIT-2106 site. (mackonstan, 14:25:47)
    4. Peter, Fan, Maciek - looks like the new latency measurements are better but still O(100usec) vs before the TRex and DPDK upgraded O(10usec). Start email thread, incl. TRex team to get to the bottom of this. (mackonstan, 14:37:52)
    5. Maciek, Tibor - will publish all new measurement data but latency. Will add rls note about ongoing issue with TRex based latency measurements; (mackonstan, 14:42:14)

  8. Releases - Maintenance CSIT-2101.1 (mackonstan, 14:44:59)
    1. wait with those tests until trex latency issue resolution, as they will be also affected (mackonstan, 14:45:41)

  9. Physical Infrastructure (mackonstan, 14:46:03)
    1. Peter, 3n-hsw decomissioning, 5 machines removed from the rack, 4 left (1of6 tasks done per vexxhost tickets). Now waiting for next tasks to be completed by Vexxhost onsite team. (mackonstan, 14:47:17)
    2. Maciek, order should have been placed, ETA is TBC. (mackonstan, 14:49:16)
    3. Juraj, arm team is actively working to provide a new Ampere Ultra server. There is a question of which NICs to equip it with. (mackonstan, 14:50:25)
    4. Maciek, the new Xeon Icelake based TGs are equipped with the following NICs: (mackonstan, 14:54:04)
    5. NIC1 Fortville Dual Port PCIe v3.0 (8.0 GT/s) 25GbE SFP+ Full Height XXV710, qty 1 per server (mackonstan, 14:54:12)
    6. NIC2 Columbiaville Quad Port PCIe v4.0 (16 GT/s x 16 lanes) 25GbE Ethernet SFP28 Intel E810XXVDA4 E810-XXVDA4, qty 1 per server (mackonstan, 14:54:17)
    7. NIC3 Columbiaville Dual Port PCIe v4.0 (16 GT/s , 2x8 lanes) 100GbE Ethernet E810-2CQDA2 (Chapman Beach), qty 1 per each server (mackonstan, 14:54:22)

  10. Virtual Infrastructure, CI/CD (mackonstan, 14:55:58)
    1. Peter, observed some JNLP resets over the last weekend, but all clear since Monday. Brought it up on vexxhost slack channel, but no acknowledgement of any RCAs. Likely due to some DC maintenance activity over the weekend. (mackonstan, 14:58:03)
    2. Vratko, Oper branches: More verification or remove?, still on TODO list (mackonstan, 14:58:44)
    3. Vratko, Scale tests: finding alternative to deprecated VAT, still on TODO list. (mackonstan, 15:00:57)
    4. Maciek, Viliam: csit-report emails - errors markup, idea is to add indexed error information to failure report emails to streamline RCAs; pls review this patch: 32898: report: indexed error message added to report emails | https://gerrit.fd.io/r/c/csit/+/32898 (mackonstan, 15:03:47)


Meeting ended at 15:03:52 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. mackonstan (38)
  2. collab-meetbot` (4)
  3. pmikus (1)
  4. fan_zhang (1)
  5. jlinkes (1)
  6. tifrank (1)
  7. vrpolak (1)


Generated by MeetBot 0.1.4.