Vratko Polak

> VPP crashes That is not confirmed. When I tried to start VPP in nodaemon mode, the CSIT code returned [2] code -1, which is a paramiko feature that means...

Different core from [3], this time on the main thread: ``` #15 0x00007ffff6d55950 in _clib_error (how_to_die=2, function_name=0x0, line_number=0, fmt=0x7ffff79ce1a7 "%s:%d (%s) assertion `%s' fails") at /w/workspace/vpp-csit-verify-perf-master-ubuntu2404-x86_64-3n-icx/src/vppinfra/error.c:143 #16 0x00007ffff7164e7f in vlib_buffer_validate...

Still present, trending [9] for CX7 on 3na-spr shows PDR more than double than MRR, and less than 200 Vectors/Call on 2n-spr [10]. [9] https://csit.fd.io/trending/#eNrdks8OgjAMxp9mXkwTGCInDyLvYeaojmSDZZsIPr2DkBQunjx52b9f-7X9Mh86h1eP-sTykhUl40VTx4Vl533cemshawV464AnyQO5TeVQ9CgCGD3koDkcDzdIJWBQ8WZFkAqMc5MGv0wa9TNsBIlYNRL5VoZShENBOZoTCehXdVbdUMjdCYO-eSPFxeaJy-gFoVRuxcNoV3QZsajmiF8ZZ-u_N24ZMRqXV7u2c2b-efHsVfeC4Bqh_fL0AVz4tlI [10] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2502-2n-spr/60/log.html.gz#s1-s1-s1-s5-s17-t1-k2-k9-k9-k14-k1-k1-k1-k1

[9] shows no change in rls2506.

[9] shows no change in rls2510.

Oh, I confused the two graphs. The two band structure is in 1c tests, and worse 2c performance is a different issue. Updating title.

Still, not sure what is causing the difference. The good run has [3] wrong (too high) Calls and Vectors counters on iavf0/0-output of vpp_wk_0, but I am not sure how...

During rls2510 this issue was overshadowed by #4109, but I suspect it is still present.

Keeping open as the trending link [0] now shows the same two bands after #4109 got fixed.

Loadbalancer tests were not covered in rls2510, but verify run [4] shows `buffer alloc error` is still present and reducing performance to zero. [4] https://logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-verify-master-2n-emr/34/log.html.gz#s1-s1-s1-s1-s1-t1-k2-k9-k10-k14-k1-k1-k1-k1