iperf3-tls/docs/dev.rst
Bruce A. Mah e142062572
Add -X to restrict SCTP binding to a subset of interfaces.
Contains an alternate implementation of previously-submitted patches
to set the maximum segment size and no-delay options.

As a result of this change, SCTP functionality on Linux will generally
require the libsctp library (on CentOS and similar distributions this
is provided by the lksctp-tools RPM).

Part of #131.

Submitted by:	Bruce Simpson <bs48@st-andrews.ac.uk>
2015-01-05 15:19:57 -08:00

238 lines
9.6 KiB
ReStructuredText

iperf3 Development
==================
The iperf3 project is hosted on GitHub at:
http://github.com/esnet/iperf
This site includes the source code repository, issue tracker, and
wiki.
Mailing Lists
-------------
The developer list for iperf3 is: iperf-dev@googlegroups.com.
Information on joining the mailing list can be found at:
http://groups.google.com/group/iperf-dev
There is, at the moment, no mailing list for user questions, although
a low volume of inquiries on the developer list is probably
acceptable. If necessary, a user-oriented mailing list might be
created in the future.
Bug Reports
-----------
Before submitting a bug report, try checking out the latest version of
the code, and confirm that it's not already fixed. Then submit to the
iperf3 issue tracker on GitHub:
https://github.com/esnet/iperf/issues
**Note:** Issues submitted to the old iperf3 issue tracker on Google
Code (or comments to existing issues on the Google Code issue tracker)
will be ignored.
Changes from iperf 2.x
----------------------
New options (not necessarily complete, please refer to the manual page
for a complete list of iperf3 options)::
-V, --verbose more detailed output than before
-J, --json output in JSON format
-Z, --zerocopy use a 'zero copy' sendfile() method of sending data
-O, --omit N omit the first n seconds (to ignore slowstart)
-T, --title str prefix every output line with this string
-F, --file name xmit/recv the specified file
-A, --affinity n/n,m set CPU affinity (Linux and FreeBSD only)
-k, --blockcount #[KMG] number of blocks (packets) to transmit (instead
of -t or -n)
-L, --flowlabel set IPv6 flow label (Linux only)
Changed flags::
-C, --linux-congestion set congestion control algorithm (Linux only)
(-Z in iperf2)
Deprecated flags (currently no plans to support)::
-d, --dualtest Do a bidirectional test simultaneously
-r, --tradeoff Do a bidirectional test individually
-T, --ttl time-to-live, for multicast (default 1)
-x, --reportexclude [CDMSV] exclude C(connection) D(data) M(multicast)
S(settings) V(server) reports
-y, --reportstyle C report as a Comma-Separated Values
Also deprecated is the ability to set the options via environment
variables.
Known Issues
------------
The following problems are notable known issues, which are probably of
interest to a large fraction of users or have high impact for some
users, and for which issues have already been filed in the issue
tracker. These issues are either open (indicating no solution
currently exists) or closed with the notation that no further attempts
to solve the problem are currently being made:
* UDP performance: Some problems have been noticed with iperf3 on the
ESnet 100G testbed at high UDP rates (above 10Gbps). The symptom is
that on any particular run of iperf3 the receiver reports a loss
rate of about 20%, regardless of the ``-b`` option used on the client
side. This problem appears not to be iperf3-specific, and may be
due to the placement of the iperf3 process on a CPU and its relation
to the inbound NIC. In some cases this problem can be mitigated by
an appropriate use of the CPU affinity (``-A``) option. (Issue #55)
* Interval reports on high-loss networks: The way iperf3 is currently
implemented, the sender write command will block until the entire
block has been written. This means that it might take several
seconds to send a full block if the network has high loss, and the
interval reports will have widely varying interval times. A
solution is being discussed, but in the meantime a work around is to
try using a small block size, for example ``-l 4K``. (Issue #125,
a fix will be released in iperf 3.1)
* The ``-Z`` flag sometimes causes the iperf3 client to hang on OSX.
(Issue #129)
* When specifying the TCP buffer size using the ``-w`` flag on Linux,
the Linux kernel automatically doubles the value passed in to
compensate for overheads. (This can be observed by using
iperf3's ``--debug`` flag.) However, CWND does not actually ramp up
to the doubled value, but only to about 75% of the doubled
value. Some part of this behavior is documented in the tcp(7)
manual page. (Issue #145)
* On some platforms (observed on at least one version of Ubuntu
Linux), it might be necessary to invoke ``ldconfig`` manually after
doing a ``make install`` before the ``iperf3`` executable can find
its shared library. (Issue #153)
There are, of course, many other open and closed issues in the issue
tracker.
Versioning
----------
iperf3 version numbers use (roughly) a `Semantic Versioning
<http://semver.org/>`_ scheme, in which version numbers consist of
three parts: *MAJOR.MINOR.PATCH*
The developers increment the:
* *MAJOR* version when making incompatible API changes,
* *MINOR* version when adding functionality in a backwards-compatible manner, and
* *PATCH* version when making backwards-compatible bug fixes.
Release Engineering Checklist
-----------------------------
1. Update the ``README`` and ``RELEASE_NOTES`` files to be accurate. Make sure
that the "Known Issues" section of the ``README`` file is up to date.
2. Compose a release announcement. Most of the release announcement
can be written before tagging. Usually the previous version's
announcement can be used as a starting point.
3. Preferably starting from a clean source tree (be sure that ``git
status`` emits no output), make the changes necessary to produce
the new version, such as bumping version numbers::
vi RELEASE_NOTES # update version number and release date
vi configure.ac # update version parameter in AC_INIT
vi src/iperf3.1 # update manpage revision date if needed
vi src/libiperf.3 # update manpage revision date if needed
git commit -a # commit changes to the local repository only
./bootstrap.sh # regenerate configure script, etc.
git commit -a # commit changes to the local repository only
# Assuming that $VERSION is the version number to be released...
./make_release tag $VERSION # this creates a tag in the local repo
./make_release tar $VERSION # create tarball and compute SHA256 hash
These steps should be done on a platform with a relatively recent
version of autotools / libtools. Examples are MacOS / MacPorts or
FreeBSD. The versions of these tools in CentOS 6 are somewhat
older and probably should be avoided.
The result will be a release artifact that should be used for
pre-testing.
4. Stage the tarball (and a file containing the SHA256 hash) to the
download site. Currently this is located on ``downloads.es.net``.
5. From another host, test the link in the release announcement by
downloading a fresh copy of the file and verifying the SHA256
checksum. Checking all other links in the release announcement is
strongly recommended as well.
6. Also verify (with file(1)) that the tarball is actually a gzipped
tarball.
7. For extra points, actually try downloading, compiling, and
smoke-testing the results of the tarball on all supported
platforms.
8. Plug the SHA256 checksum into the release announcement.
9. PGP-sign the release announcement text using ``gpg --clearsign``.
The signed announcement will be sent out in a subsequent emails,
but could also be archived. Decoupling the signing from emailing
allows a signed release announcement to be resent via email or sent
by other, non-email means.
10. At this point, the release can and should be considered
finalized. To commit the release-engineering-related changes to
GitHub and make them public, push them out thusly::
git push # Push version changes
git push --tags # Push the new tag to the GitHub repo
11. Send the PGP-signed release announcement to the following
addresses. Remember to turn off signing in the MUA, if
applicable. Remember to check the source address when posting to
lists, as "closed" list will reject posting from all from
registered email addresses.
* iperf-dev@googlegroups.com
* iperf-users@lists.sourceforge.net
* perfsonar-user@internet2.edu
* perfsonar-developer@internet2.edu
Note: Thunderbird sometimes mangles the PGP-signed release
announcement so that it does not verify correctly. This could be
due to Thunderbird trying to wrap the length of extremely long
lines (such as the SHA256 hash). Apple Mail and mutt seem to
handle this situation correctly. Testing the release announcement
sending process by sending a copy to oneself first and attempting
to verify the signature is highly encouraged.
12. Update the iperf3 Project News section of the documentation site
to announce the new release (see ``docs/news.rst`` and
``docs/conf.py`` in the source tree) and deploy a new build of the
documentation to GitHub Pages.
Code Authors
------------
The main authors of iperf3 are (in alphabetical order): Jon Dugan,
Seth Elliott, Bruce A. Mah, Jeff Poskanzer, Kaustubh Prabhu.
Additional code contributions have come from (also in alphabetical
order): Mark Ashley, Aaron Brown, Aeneas Jaißle, Susant Sahani,
Bruce Simpson, Brian Tierney.
iperf3 contains some original code from iperf2. The authors of iperf2
are (in alphabetical order): Jon Dugan, John Estabrook, Jim Ferbuson,
Andrew Gallatin, Mark Gates, Kevin Gibbs, Stephen Hemminger, Nathan
Jones, Feng Qin, Gerrit Renker, Ajay Tirumala, Alex Warshavsky.