Regression in --txstart-time
A means to measure network responsiveness and throughput
Brought to you by:
rjmcmahon
Hi Bob,
With new iperf we see client connecting starts after specified txstart time
16:29:30.177 LOG 10.123.6.191 iperf -v
16:29:30.211 LOG 10.123.6.191 iperf version 2.1.n (5 June 2023) pthreads
16:19:21.271 DEBUG IPERFCMD iperf -c 192.168.1.195 -e -w 24M -B 192.168.1.218 --bounceback --bounceback-hold 0 --bounceback-period 1 --hide-ips -l 100 --trip-times --txstart-time 1687862969 -i 1 -fb -S 0x00 -T 255 -t 30 -p 61001 --trip-times --histogram
16:19:21.272 INFO r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx calling rpopen: iperf -c 192.168.1.195 -e -w 24M -B 192.168.1.218 --bounceback --bounceback-hold 0 --bounceback-period 1 --hide-ips -l 100 --trip-times --txstart-time 1687862969 -i 1 -fb -S 0x00 -T 255 -t 30 -p 61001 --trip-times --histogram
16:19:21.273 LOG 10.123.6.191 iperf -c 192.168.1.195 -e -w 24M -B 192.168.1.218 --bounceback --bounceback-hold 0 --bounceback-period 1 --hide-ips -l 100 --trip-times --txstart-time 1687862969 -i 1 -fb -S 0x00 -T 255 -t 30 -p 61001 --trip-times --histogram
16:19:21.284 VWAIT r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx vwait on READY : state is INIT
16:19:21.300 LOG 4916> CONSOLE[wl0]: 402510.644 wl0: wlc_iovar_ext: dump_clear: BCME -23
16:19:23.518 LOG 10.123.6.195> CONSOLE_E:1:489 4100.201476375 noise_cal_info: scan core: params: 0 1e c8,caller: 163647
16:19:23.520 LOG 10.123.6.195> CONSOLE_E:1:490 4100.203003029 noise_cal_info: scan core: params: 0 1e c8,caller: 163647
16:19:24.284 ERROR r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx iperf tx start timeout
In streams lib we look streams successfully connected using "regexp {^Client connecting to} $buf". Observe timeout issue in streams lib with new iperf as client connecting starts after specified txstart time.
With old iperf we see client connecting once we trigger the iperf client. But data transfer starts only after "txstart start time specified"
[root@bca14end1 ~]# iperf8m --v
iperf version 2.1.n (8 March 2023) pthreads
16:29:38.789 DEBUG IPERFCMD iperf8m -c 192.168.1.195 -e -w 24M -B 192.168.1.218 --bounceback --bounceback-hold 0 --bounceback-period 1 --hide-ips -l 100 --trip-times --txstart-time 1687863586 -i 1 -fb -S 0x00 -T 255 -t 30 -p 61001 --trip-times --histogram
16:29:38.789 INFO r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx calling rpopen: iperf8m -c 192.168.1.195 -e -w 24M -B 192.168.1.218 --bounceback --bounceback-hold 0 --bounceback-period 1 --hide-ips -l 100 --trip-times --txstart-time 1687863586 -i 1 -fb -S 0x00 -T 255 -t 30 -p 61001 --trip-times --histogram
16:29:38.791 LOG 10.123.6.191 iperf8m -c 192.168.1.195 -e -w 24M -B 192.168.1.218 --bounceback --bounceback-hold 0 --bounceback-period 1 --hide-ips -l 100 --trip-times --txstart-time 1687863586 -i 1 -fb -S 0x00 -T 255 -t 30 -p 61001 --trip-times --histogram
16:29:38.800 VWAIT r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx vwait on READY : state is INIT
16:29:38.801 LOG 4916> CONSOLE[wl0]: 403128.385 wl0: wlc_iovar_ext: dump_clear: BCME -23
16:29:38.812 HNDLR r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx ------------------------------------------------------------ (lan,file18)
16:29:38.813 HNDLR r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx Client connecting to (**hidden**), TCP port 61001 with pid 12806 (1 flows) (lan,file18)
16:29:38.813 HNDLR r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx Handler now SAMPLING (lan,file18)
16:29:38.813 VWAIT r6726_5G_TAG_LLAPI-4398_STA1-DOWN-1-tx vwait DONE : state is READY
Regards,
Guru
It looks like a regression. Let me check into it. The design was not to delay the TCP 3WHS (connect) but only delay the data xfer phase by the --txstart-time.