perfsonar-user - Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3
Subject: perfSONAR User Q&A and Other Discussion
List archive
- From: Brian Candler <>
- To: Aaron Brown <>
- Cc: "" <>
- Subject: Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3
- Date: Wed, 10 Dec 2014 21:32:50 +0000
- Domainkey-signature: a=rsa-sha1; c=nofws; d=pobox.com; h=message-id:date :from:mime-version:to:cc:subject:references:in-reply-to :content-type:content-transfer-encoding; q=dns; s=sasl; b=pZaOnZ TsMMlx7Gn4GlU8EFhAu9Ppsr3rUCfTrLqu3JlclA1LaAS81Zs3aghhyAsl2KVfqF qGa97ayYvava/hIx4hFEL8Z/CoRWPyaUQmf2WTcuz3mjzcEU7sr8/2HMTSNMf+mL NvyMhKvDVejTXD0r97OWtZbTtc8ZFZcRIua3c=
On 10/12/2014 14:23, Aaron Brown wrote:
Could you try applying thisThis one also doesn't give any output (see end of mail).
patch:http://ndb1.internet2.edu/~aaron/bwctl_change_iperf3_exec.patch (or
just downloadhttp://ndb1.internet2.edu/~aaron/iperf3.c and replace the
iperf3.c in ‘bwlib’). This changes it so that bwctl just execs iperf3 instead
of linking to the library.
I know it's the patched daemon I'm running because it includes
bwctld[13564]: FILE=iperf3.c, LINE=65, Iperf3Available(): iperf3_cmd unset, using "iperf3"
in its startup message.
If I "strace -f" the parent bwctld process at the other side (cov-perf-1) I see:
# grep exec /var/tmp/bwctl.trace
[pid 55003] write(1, "bwctl: exec_line: iperf3 -c 212."..., 238) = 238
[pid 55003] execve("/sbin/iperf3", ["iperf3", "-c", "192.0.2.251", "-B", "192.0.2.254", "-f", "m", "-p", "5792", "-V", "-t", "10"], [/* 5 vars */]) = -1 ENOENT (No such file or directory)
[pid 55003] execve("/usr/sbin/iperf3", ["iperf3", "-c", "192.0.2.251", "-B", "192.0.2.254", "-f", "m", "-p", "5792", "-V", "-t", "10"], [/* 5 vars */]) = -1 ENOENT (No such file or directory)
[pid 55003] execve("/bin/iperf3", ["iperf3", "-c", "192.0.2.251", "-B", "192.0.2.254", "-f", "m", "-p", "5792", "-V", "-t", "10"], [/* 5 vars */]) = -1 ENOENT (No such file or directory)
[pid 55003] execve("/usr/bin/iperf3", ["iperf3", "-c", "192.0.2.251", "-B", "192.0.2.254", "-f", "m", "-p", "5792", "-V", "-t", "10"], [/* 5 vars */]) = 0
[pid 55006] execve("/bin/sh", ["sh", "-c", "uname -a"], [/* 5 vars */]) = 0
[pid 55006] execve("/bin/uname", ["uname", "-a"], [/* 5 vars */]) = 0
[pid 55002] write(13, ".254\nbwctl: exec", 16) = 16
[pid 55002] read(13, ".254\nbwctl: exec", 16) = 16
[pid 55001] write(10, ".254\nbwctl: exec", 16) = 16
[pid 55011] write(1, "bwctl: nothing to exec for trace"..., 199) = 199
[pid 55010] write(14, "ing to exec for ", 16) = 16
Grepping for process 55003 I see:
...
[pid 55003] socket(PF_INET, SOCK_STREAM, IPPROTO_IP) = 8
[pid 55003] bind(8, {sa_family=AF_INET, sin_port=htons(0), sin_addr=inet_addr("192.0.2.254")}, 16) = 0
[pid 55003] connect(8, {sa_family=AF_INET, sin_port=htons(5792), sin_addr=inet_addr("192.0.2.251")}, 16) = -1 ECONNREFUSED (Connection refused)
[pid 55003] close(8) = 0
[pid 55003] write(2, "iperf3: error - unable to connec"..., 64) = 64
[pid 55003] exit_group(1) = ?
So it looks like the iperf3 -s simply isn't ready in time. Simple race condition? How much slack does bwctld allow between starting the server and starting the client?
Regards,
Brian.
[root@bou-perf-1
~]# time bwctl -v -x -a 2 -s cov-perf-1 -T iperf3
Messages being sent to syslog(user,err)
bwctl: Using cov-perf-1 as the address for remote sender
bwctl: Using 192.0.2.251 as the address for local receiver
bwctl: Available in-common: iperf nuttcp iperf3
bwctl: Using tool: iperf3
bwctl: Server 'localhost' accepted test request at time 1418246840.702620
bwctl: Client 'cov-perf-1' accepted test request at time 1418246840.702620
bwctl: 43 seconds until test results available
RECEIVER START
bwctl: start_endpoint: 3627235608.629568
bwctl: run_endpoint: receiver: 192.0.2.251
bwctl: run_endpoint: sender: 192.0.2.254
bwctl: exec_line: iperf3 -s -B 192.0.2.251 -f m -p 5792 -V
bwctl: run_tool: tester: iperf3
bwctl: run_tool: receiver: 192.0.2.251
bwctl: run_tool: sender: 192.0.2.254
bwctl: start_tool: 3627235625.162757
iperf 3.0.9
Linux bou-perf-1.network.example.com 2.6.32-504.1.3.el6.aufs.web100.x86_64 #1 SMP Wed Nov 12 16:01:48 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
iperf3: interrupt - the server has terminated
-----------------------------------------------------------
Server listening on 5792
-----------------------------------------------------------
bwctl: stop_tool: 3627235656.684916
bwctl: stop_endpoint: 3627235656.685499
RECEIVER END
SENDER START
bwctl: start_endpoint: 3627235608.676979
bwctl: run_endpoint: receiver: 192.0.2.251
bwctl: run_endpoint: sender: 192.0.2.254
bwctl: exec_line: iperf3 -c 192.0.2.251 -B 192.0.2.254 -f m -p 5792 -V -t 10
bwctl: run_tool: tester: iperf3
bwctl: run_tool: receiver: 192.0.2.251
bwctl: run_tool: sender: 192.0.2.254
bwctl: start_tool: 3627235640.703108
iperf 3.0.9
Linux cov-perf-1.network.example.com 2.6.32-504.1.3.el6.aufs.web100.x86_64 #1 SMP Wed Nov 12 16:01:48 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
iperf3: error - unable to connect to server: Connection refused
bwctl: stop_tool: 3627235640.728257
bwctl: stop_endpoint: 3627235656.735306
SENDER END
real 0m48.170s
user 0m0.002s
sys 0m0.004s
[root@bou-perf-1
~]#
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, (continued)
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/05/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/05/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/06/2014
- RE: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Tim Rayner, 12/06/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/07/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/07/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Aaron Brown, 12/09/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/09/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/09/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Aaron Brown, 12/10/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/10/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/11/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Aaron Brown, 12/11/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/11/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/07/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/07/2014
- RE: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Tim Rayner, 12/06/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/10/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/11/2014
- Re: [perfsonar-user] Different results from iperf3 vs. bwctl -T iperf3, Brian Candler, 12/06/2014
Archive powered by MHonArc 2.6.16.