[CSIT-1941] TRex may wrongly detect link bandwidth

[vvalderrv]

Description

This issue is somewhat similar to CSIT-1936 but happens very rarely. Or perhaps it only rarely leads to test failure, as 64B test cases are likely to pass even with wrong bandwidth.

One occurrence [0], only one test case affected (Intel-x520 and dpdk plugin):

trex.common.trex_exceptions.TRexError: Port 0 : *** Expected L1 B/W: '8.73783 Gbps' exceeds port line rate: '1 Gbps'

Another occurrence [1], all test cases in the suite are affected (Intel-e810cq and AVF plugin):

trex.common.trex_exceptions.TRexError: Port 0 : *** Expected L1 B/W: '96.608 Gbps' exceeds port line rate: '50 Gbps'

[0] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2310-3n-tsh/20/log.html.gz#s1-s1-s1-s1-s6-t5-k2-k13-k9-k10-k1-k1-k1-k11

[1] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2310-3nb-spr/28/log.html.gz#s1-s1-s1-s1-s4-t1-k2-k13-k14

Assignee

Unassigned

Reporter

Vratko Polak

Comments

• vrpolak (Tue, 19 Nov 2024 12:38:59 +0000): The first value depends on traffic used, e.g. imix has 98.9413 Gbps [3] in rls2410.

[3] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-coverage-2410-3nb-spr/32/log.html.gz#s1-s1-s1-s1-s1-t7-k2-k9-k14

• vrpolak (Mon, 29 Jul 2024 08:27:46 +0000): Rare, but still possible [2] in rls2406.

[2] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2406-3n-tsh/41/log.html.gz#s1-s1-s1-s5-s1-t3-k2-k9-k9-k10-k1-k1-k1-k11

Original issue: https://jira.fd.io/browse/CSIT-1941

[vvalderrv]

The first value depends on traffic used, e.g. imix has 98.9413 Gbps [3] in rls2410.

[3] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-coverage-2410-3nb-spr/32/log.html.gz#s1-s1-s1-s1-s1-t7-k2-k9-k14

[vvalderrv]

Rare, but still possible [2] in rls2406.

[2] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2406-3n-tsh/41/log.html.gz#s1-s1-s1-s5-s1-t3-k2-k9-k9-k10-k1-k1-k1-k11

[vrpolakatcisco]

Still present in rls2502, IMIX traffic in ipsec tests attempts different date (due to encapsulation overhead), but still fails if 100GE is misdetected as 50GE [4]:

trex.common.trex_exceptions.TRexError: Port 1 : *** Expected L1 B/W: '82.737 Gbps' exceeds port line rate: '50 Gbps'

[4] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2502-3nb-spr/49/log.html.gz#s1-s1-s1-s1-s7-t4-k2-k10-k14

[vrpolakatcisco]

Still present [5] in rls2506.

[5] https://logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2506-2n-spr/72/log.html.gz#s1-s1-s1-s1-s1-t2-k2-k10-k13-k9

[vrpolakatcisco]

Still present [6] in rls2510. It is suspicious that other 3-node testbeds do not show this symptom.

[6] https://logs.fd.io/vex-yul-rot-jenkins-1/csit-vpp-perf-report-iterative-2510-3nb-spr/38/log.html.gz#s1-s1-s1-s1-s1-t1-k2-k10-k15-k9