=========================================== #fdio-meeting: "FD.io CSIT Project Meeting" =========================================== Meeting started by mackonstan at 15:03:16 UTC. The full logs are available at http://ircbot.wl.linuxfoundation.org/meetings/fdio-meeting/2024/_fd_io_csit_project_meeting_/fdio-meeting-_fd_io_csit_project_meeting_.2024-01-17-15.03.log.html . Meeting summary --------------- * Agenda bashing (mackonstan, 15:03:27) * Vratko Polak. (vrpolak, 15:03:38) * Tibor Frank (tifrank, 15:03:51) * VPP update (mackonstan, 15:06:07) * Andrew 1 wk away from rc1 (mackonstan, 15:06:32) * TSC update (mackonstan, 15:06:50) * HW accelerator - SPR QAT2.0 (mackonstan, 15:08:52) * copied notes from last week: Kai - issues reported that QAT stopped working, intel team determined that tests are using the same VF for QAT ingress and QAT egress interfaces. and they should be using a different VF for QAT ingress and a different one for QAT egress interfaces. See separate email thread on this (mackonstan, 15:09:28) * Dave Wallace (dwallacelf, 15:09:54) * Kai - synced up with Peter before the meeting - root cause is that single QAT VF is doing both encrypt and decrypt - conclusion is that the test case is not optimized with the current (latest) QAT HW setup. (mackonstan, 15:11:29) * Kai - plan is to modify the test cases - Kai's Intel team will do the work with Peter's guidance. (mackonstan, 15:12:53) * HW accelerator - SNR QAT2.0 (mackonstan, 15:13:12) * notes from last week: Kai - after processor replacement the tests should be good to run now, as the configuration is the same as in Intel shadow lab. Peter to confirm. (mackonstan, 15:13:29) * On TSC meeting, there was a discussion about using openprofile.dev for tracking meetings (on LF level). But most projects use Google calendars with strong opinions against migrating to anything worse. (vrpolak, 15:14:03) * Kai - base ip4 and ip6 tests are running. for QAT2.0 ipsec async tests will need modified as for SPR QAT - the same team and work arrangement as for SPR QAT. (mackonstan, 15:14:49) * HW accelerator - SPR DSA (mackonstan, 15:15:03) * notes from last week: Kai - existing issue with intel idxd driver - Intel team confirms that to run this driver it requires newer kernel version. current ver. 5.15, required ver. 5.18+. Peter to advise once back from PTO. (mackonstan, 15:15:21) * Kai - good news and bad news. (mackonstan, 15:15:32) * Kai - bad news - need to upgrade to kernel 5.18+ - bigger effort. (mackonstan, 15:17:28) * - good news - some parameters in kernel to disable level-5 paging in intel cpus (using grub command) - may solve the issue with idxd driver. if Peter can give it a try to check if it will work. (mackonstan, 15:18:28) * Peter - will try the grub cmd approach first, if it doesn't work, then the next option will be to upgrade kernel, but may break compatibility. (mackonstan, 15:19:47) * Physical Infrastructure - 1n-alt onboarding (mackonstan, 15:21:13) * Dave asking about status, as no info was coming from Juraj (mackonstan, 15:21:25) * Peter confirmed that coordination is already ongoing with Vexxhost for installation and bringing both 1n-alt rigs online (mackonstan, 15:21:51) * Peter - should be up by eow (mackonstan, 15:22:00) * Peter working with Juraj to make it work (mackonstan, 15:22:14) * Dave - will push a patch ci-mgmt for disabling 1n-tx2 (mackonstan, 15:23:15) * Dave - in the background Naveen is to work in debugging 1n-spr vpp_device jobs that are still not voting due to reliability issues (mackonstan, 15:24:42) * Peter - 1n-spr vpp_device jobs are now more stable, more in the greenish zone, but not fully out of the woods yet (mackonstan, 15:25:26) * CSIT-2402 release (mackonstan, 15:25:47) * Tibor - ready for release (mackonstan, 15:28:16) * great improvement - 40-50% execution time reduction (mackonstan, 15:30:50) * LINK: https://csit.fd.io/stats/#eNrLyk-yTS7OLNEtKyjQLUgtStPNSykqSCnSLU9Nzc6p1M1NLC5JLdI1ytPNTK4AAIoqEPU (mackonstan, 15:31:00) * LINK: yTS7OLNEtKyjQLUgtStPNSykqSCnSLU9Nzc6p1M1NLC5JLdI1ytNNzqkAAIoqEPg (mackonstan, 15:31:15) * LINK: https://csit.fd.io/stats/#eNrLyk-yTS7OLNEtKyjQLUgtStPNSykqSCnSLU9Nzc6p1M1NLC5JLdI1ytNNzqkAAIoqEPg (mackonstan, 15:31:25) * LINK: https://csit.fd.io/stats/#eNrLyk-yTS7OLNEtKyjQLUgtStPNLSrSTUnMzKnUzU0sLkkt0jXK0y0uKAIARFMPTw (mackonstan, 15:31:40) * Vratko Maciek Dave discuss how to share the value of MLRsearch for volume continuous execution of RFC2544-like benchmarking tests with wider community (mackonstan, 15:34:33) * Environment, Framework (mackonstan, 15:35:32) * Vratko - for next week need to update the api crc values we are using (mackonstan, 15:35:51) * Dave - new ubuntu 24.04 coming out in April. started preliminary work to start planning the staging work to happen ... (mackonstan, 15:37:25) * Trending and Test Failures (mackonstan, 15:38:50) * Vratko - wiki pages not updated (mackonstan, 15:38:57) * Vratko - dashboard trending pages error rate hasn't changed for a while (mackonstan, 15:39:19) * Vratko - srv6 failures, duplicate root cause - see jira ticket (mackonstan, 15:40:11) * Vratko - CSIT-1945 https://jira.fd.io/browse/CSIT-1945 Some srv6 9000B tests crash VPP (mackonstan, 15:41:08) * LINK: https://jira.fd.io/browse/CSIT-1945 (mackonstan, 15:41:21) * Vratko - another new ticket CSIT-1946 ipsec hwasync fails with large scale and multiple queues (mackonstan, 15:41:58) * LINK: https://jira.fd.io/browse/CSIT-1946 (mackonstan, 15:42:07) * Fd.io Events (mackonstan, 15:42:19) * Dave - f2f meeting - open source summit and lfn ddtf - 29apr-3may (mackonstan, 15:43:20) * Dave - see TSC wiki pages for event details (mackonstan, 15:43:32) * LINK: https://wiki.fd.io/view/TSC (mackonstan, 15:44:06) * Maciek - can't make meeting next week, Tibor kindly agreed to lead the csit call (mackonstan, 15:44:56) Meeting ended at 15:44:59 UTC. People present (lines said) --------------------------- * mackonstan (52) * collab-meetbot (10) * vrpolak (2) * dwallacelf (1) * tifrank (1) * jlinkes (0) * Kai17 (0) Generated by `MeetBot`_ 0.1.4