Benchmark test not executing due to tsr_error

[francovz96]

Description, req_size, resp_size, rate, tx pps, rx pps, tx usage, rx usage Test case didn't pass: tsr_error { e_code: 0 } tsr_state: RUNNING tsr_type: CLIENT tsr_l4_proto: TCP tsr_app_proto: RAW_CLIENT tsr_stats { gs_up: 9572509 gs_estab: 9572497 gs_down: 0 gs_failed: 0 gs_data_failed: 0 gs_data_null: 0 gs_start_time: 619591557837 gs_end_time: 619595547889 gs_latency_stats { gls_stats { ls_max_exceeded: 0 ls_max_average_exceeded: 0 ls_min_latency: 0 ls_max_latency: 0 ls_sum_latency: 0 ls_sum_jitter: 0 ls_instant_jitter: 0 ls_samples_count: 0 } gls_sample_stats { ls_max_exceeded: 0 ls_max_average_exceeded: 0 ls_min_latency: 0 ls_max_latency: 0 ls_sum_latency: 0 ls_sum_jitter: 0 ls_instant_jitter: 0 ls_samples_count: 0 } gls_invalid_lat: 0 } } tsr_rate_stats { rs_estab_per_s: 1159403 rs_closed_per_s: 0 rs_data_per_s: 0 rs_start_time: 619594547930 rs_end_time: 619595547902 } tsr_app_stats { as_raw { rsts_req_cnt: 0 rsts_resp_cnt: 0 rsts_tstamp_no_phys_addr: 0 rsts_tstamp_no_mbuf: 0 rsts_tstamp_no_win: 0 } } Should cleanup but we just exit..
With the ini file
test_2_perf_benchmark.txt

[davvore33]

Hi @francovz96
I see here tsr_error is 0 which is how it is supposed to be, can I know how did you collect this message?

[francovz96]

Hi @davvore33,

I collect this meassage when running the python test_2_perf_benchmark.py, with the configuration in the test_2_perf_benchmark.txt file above. The test does not run and does not collect any results in benchmark.csv.

[davvore33]

Hi @francovz96

I've just tried now and here it works, could you please paste the full command line you run?

You can try to raise warp17 directly from the command line like this
./build/warp17 -c 0xFC3 -n 4 -m 24576 -w 65:00.0 -w 65:00.1 -- --tcb-pool-sz 24576 --ucb-pool-sz 24576 --qmap 65:00.0,0x4 --qmap 65:00.1,0x8
and than run the benchmark example so it will "send" the test configurations to warp17 and you will be able to see if there are any errors

[francovz96]

Hi @davvore33,

thanks very much for the quick reply, the command I run is : python -u test_2_perf_benchmark.py -f /root/warp17/build/warp17 > ../../benchmarks/benchmark.csv . I will try to run the example you gave.

[francovz96]

Hi @davvore33,

When I run warp17(top left), python command (bottom left) and test_2_perf.log(top right), this is the output I get, CPUS are still running but python scripts has been exited and nothin is happening on warp17 screen, could you maybe attach the test_2_perf_benchmark.py file you are using so that I can compare ?

[francovz96]

Hi @davvore33 ,

I have found the place in the test_2_perf_benchmark.py file where it exits to be as in the figure below. Looks like result.tsr_state = FAILED, what could cause this ?

[davvore33]

hi @francovz96,
so we spawn as many threads as many cores you give to warp17 and those threads keep looping forever even without any test configured, have said that, reason to FAIL are many,
on the warp17 cli, in your first screen top left, you can try to figure out if there's something wrong,
i suggest you to dig using the following commands in order:

• check for the arp table
  show arp entries
• check if the traffic is flowing and if it has errors

show ipv4 statistics
show tcp statistics #I'm not sure if it was http/raw on tcp or udp

in case you spot some tcp errors, you can even check for the tcp state machine stats
show tsm statistics

let me know how is going

[francovz96]

Hi @davvore33,

Looking at the arp entries after test_2_perf_benchmark.py stops running I find this

The arp statistics look like this,

The ipv4 statistics look as follows,

The tcp statistics look as follows,

The TSM statistics looks as follows,

Please let me know if any of this shows where it is breaking or if I can send you any more information