crypto engine error in 1tnl wireguardhw test

[vrpolakatcisco]

I am sure I have seen "crypto engine error (packet dropped)" already, but I not could not find such a comment in any issue. Maybe I have seen that when investigating something unrelated, maybe my search on GitHub is not good (and this ticket is a duplicate).

I do not believe this is a symptom of #3968, so it needs to be tracked separately. Trending is not useful at tracking this issue, as TRex bugs such as #4018 and #4023 were more frequent in rls2506. Maybe this issue is caused by ubuntu2404 migration, maybe it is older but misclassified or rare.

In this test all packets were lost, with the error counter visible in teardown stats [0].

[0] https://logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2506-3n-emr/12/log.html.gz#s1-s1-s1-s3-s10-t1-k3-k7-k1-k1-k1-k8-k14-k1-k1-k1-k1

[vrpolakatcisco]

Another run [1] on a different testbed. Note that #4045 can affect the same tests, the difference is whether VPP stays running until teardown.

[1] https://logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2506-3nb-spr/19/log.html.gz#s1-s1-s1-s3-s10-t2-k3-k7-k1-k1-k1-k8-k14-k1-k1-k1-k1

[vrpolakatcisco]

Still present [2] in rls2510.

[2] https://logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2510-3nb-spr/29/log.html.gz#s1-s1-s1-s3-s10-t5-k3-k7-k1-k1-k1-k8-k14-k1-k1-k1-k1