Alkis Georgopoulos

Re: date issue, I just tested building from master. The date is correctly reported there: alkisg@alkis:~/tmp/iperf2-code$ src/iperf -v iperf version 2.1.n (9 Jan 2023) pthreads alkisg@alkis:~/tmp/iperf2-code$ src/iperf -s -xS -yC 20230121222428,127.0.0.1,51228,127.0.0.1,5001,1,0.0-10.0,55389454396,44309600601 20230121222438,127.0.0.1,44586,127.0.0.1,5001,3,0.0-10.0,55052599292,44041846011 20230121222438,127.0.0.1,44586,127.0.0.1,5001,2,0.0-10.0,55052599296,43972067111 But then I tried downloading...

Re: id feature request, thanks, continuing there. Re: date issue, sure I can test building from master, but please allow me a few days to finish some other urgent tasks first.

I would vote for passing an ID, as it would result in very readable output: 19700101020000.000,127.0.0.1,33952,127.0.0.1,5001,1,0.0-10.0,53973876796,43177974491,pc01 19700101020000.000,127.0.0.1,55512,127.0.0.1,5001,2,0.0-10.0,53349843004,42678713542,pc02 The id column could always exist, it could default to the hostname, and it could be customizable with the --id=xxx parameter. I don't think that passing both the internal and external ip/ports would be preferable, because the output would be rather...

Re: date, it happens with all the version I've tested, e.g. 2.1.5, 2.1.7, 2.1.8, in all Debian derivatives like Debian, Ubuntu, Raspberry Pi OS, Proxmox etc. Edit, I just tested and it also happens with iperf 2.1.8 on Fedora 36, i.e. I haven't seen anything other than 1970 in any of my tests. Re: port, the problem with letting the user assign unique port numbers is that it then becomes a difficult "cluster" issue; the user will have to develop a program that will cooperate with all the "cluster"...

Re: date, it happens with all the version I've tested, e.g. 2.1.5, 2.1.7, 2.1.8, in all Debian derivatives like Debian, Ubuntu, Raspberry Pi OS, Proxmox etc. Edit, I just tested and it also happens with iperf 2.1.8 on Fedora 36, i.e. I haven't see anything else than 1970 in any of my tests. Re: port, the problem with letting the user assign unique port numbers is that it then becomes a difficult "cluster" issue; the user will have to develop a program that will cooperate with all the "cluster" nodes...

Re: date, it happens with all the version I've tested, e.g. 2.1.5, 2.1.7, 2.1.8, in all Debian derivatives like Debian, Ubuntu, Raspberry Pi OS, Proxmox etc. Re: port, the problem with letting the user assign unique port numbers is that it then becomes a difficult "cluster" issue; the user will have to develop a program that will cooperate with all the "cluster" nodes (the PCs that participate in the iperf benchmark), try to reserve some specific ports, report back to the "coordinating server", see...

Re: date, it happens with all the version I've tested, e.g. 2.1.5, 2.1.7, 2.1.8. Re: port, the problem with letting the user assign unique port numbers is that it then becomes a difficult "cluster" issue; the user will have to develop a program that will cooperate with all the "cluster" nodes (the PCs that participate in the iperf benchmark), try to reserve some specific ports, report back to the "coordinating server", see which ports are in use or duplicate, propose new ports, and iterate until...

Re: date, it happens with all the version I've tested, e.g. 2.1.5, 2.1.7, 2.1.8. Re: port, the problem with letting the user assign unique port numbers is that it then becomes a difficult "cluster" issue; the user will have to develop a program that will cooperate with all the "cluster" nodes (the PCs that participate in the iperf benchmark), try to reserve some specific ports, report back to the "coordinating server", see which ports are in use or duplicate, propose new ports, and iterate until...