<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Hi,<br>
<br>
Please find some answers inline.<br>
<br>
Regards,<br>
Alexei<br>
<br>
On 06/10/2014 03:06 PM, Flavio Percoco wrote:<br>
</div>
<blockquote cite="mid:20140610120649.GG20684@redhat.com" type="cite">On
10/06/14 15:03 +0400, Dina Belova wrote:
<br>
<blockquote type="cite">Hello, stackers!
<br>
<br>
<br>
Oslo.messaging is future of how different OpenStack components
communicate with
<br>
each other, and really I’d love to start discussion about how we
can make this
<br>
library even better then it’s now and how can we refactor it
make more
<br>
production-ready.
<br>
<br>
<br>
As we all remember, oslo.messaging was initially inspired to be
created as a
<br>
logical continuation of nova.rpc - as a separated library, with
lots of
<br>
transports supported, etc. That’s why oslo.messaging inherited
not only
<br>
advantages of now did the nova.rpc work (and it were lots of
them), but also
<br>
some architectural decisions that currently sometimes lead to
the performance
<br>
issues (we met some of them while Ceilometer performance testing
[1] during the
<br>
Icehouse).
<br>
<br>
<br>
For instance, simple testing messaging server (with connection
pool and
<br>
eventlet) can process 700 messages per second. The same
functionality
<br>
implemented using plain kombu (without connection pool and
eventlet) driver is
<br>
processing ten times more - 7000-8000 messages per second.
<br>
<br>
<br>
So we have the following suggestions about how we may make this
process better
<br>
and quicker (and really I’d love to collect your feedback,
folks):
<br>
<br>
<br>
1) Currently we have main loop running in the Executor class,
and I guess it’ll
<br>
be much better to move it to the Server class, as it’ll make
relationship
<br>
between the classes easier and will leave Executor only one task
- process the
<br>
message and that’s it (in blocking or eventlet mode). Moreover,
this will make
<br>
further refactoring much easier.
<br>
</blockquote>
<br>
To some extent, the executors are part of the server class since
the
<br>
later is the one actually controlling them. If I understood your
<br>
proposal, the server class would implement the event loop, which
means
<br>
we would have an EventletServer / BlockingServer, right?
<br>
<br>
If what I said is what you meant, then I disagree. Executors keep
the
<br>
eventloop isolated from other parts of the library and this is
really
<br>
important for us. One of the reason is to easily support multiple
<br>
python versions - by having different event loops.
<br>
<br>
Is my assumption correct? Could you elaborate more?
<br>
</blockquote>
No It's not how we plan it. Server will do the loop and pass
received message to dispatcher and executor. It means that we would
still have blocking executor and eventlet executor in the same
server class. We would just change the implementation part to make
it more consistent and easier to control.<br>
<br>
<blockquote cite="mid:20140610120649.GG20684@redhat.com" type="cite">
<br>
<blockquote type="cite">
<br>
2) Some of the drivers implementations (such as impl_rabbit and
impl_qpid, for
<br>
instance) are full of useless separated classes that in reality
might be
<br>
included to other ones. There are already some changes making
the whole
<br>
structure easier [2], and after the 1st issue will be solved
Dispatcher and
<br>
Listener also will be able to be refactored.
<br>
</blockquote>
<br>
This was done on purpose. The idea was to focus on backwards
<br>
compatibility rather than cleaning up/improving the drivers. That
<br>
said, sounds like those drivers could user some clean up. However,
I
<br>
think we should first extend the test suite a bit more before
hacking
<br>
the existing drivers.
<br>
<br>
<br>
<blockquote type="cite">
<br>
3) If we’ll separate RPC functionality and messaging
functionality it’ll make
<br>
code base clean and easily reused.
<br>
</blockquote>
<br>
What do you mean with this?
<br>
</blockquote>
We mean that current drivers are written with RPC code hardcoded
inside (ReplyWaiter, etc.). Thats not how messaging library is
supposed to work. We can move RPC to a separate layer and this would
be beneficial for both rpc (code will become more clean and less
error-prone) and core messaging part (we'll be able to implement
messaging in way that will work much faster).<br>
<blockquote cite="mid:20140610120649.GG20684@redhat.com" type="cite">
<br>
<blockquote type="cite">
<br>
4) Connection pool can be refactored to implement more efficient
connection
<br>
reusage.
<br>
</blockquote>
<br>
Please, elaborate. What changes do you envision?
<br>
</blockquote>
Currently there is a class that is called ConnectionContext that is
used to manage pool. Additionaly it can be accessed/configured in
several other places. If we refactor it a little bit it would be
much easier to use connections from the pool.<br>
<blockquote cite="mid:20140610120649.GG20684@redhat.com" type="cite">
<br>
As Dims suggested, I think filing some specs for this (and keeping
the
<br>
proposals separate) would help a lot in understanding what the
exact
<br>
plan is.
<br>
<br>
Glad to know you're looking forward to help improving
oslo.messaging.
<br>
<br>
Thanks,
<br>
Flavio
<br>
<br>
<blockquote type="cite">Folks, are you ok with such a plan? Alexey
Kornienko already started some of
<br>
this work [2], but really we want to be sure that we chose the
correct vector
<br>
of development here.
<br>
<br>
<br>
Thanks!
<br>
<br>
<br>
[1] <a class="moz-txt-link-freetext" href="https://docs.google.com/document/d/">https://docs.google.com/document/d/</a>
<br>
1ARpKiYW2WN94JloG0prNcLjMeom-ySVhe8fvjXG_uRU/edit?usp=sharing
<br>
<br>
[2] <a class="moz-txt-link-freetext" href="https://review.openstack.org/#/q/">https://review.openstack.org/#/q/</a>
<br>
status:open+owner:akornienko+project:openstack/oslo.messaging,n,z
<br>
<br>
<br>
Best regards,
<br>
<br>
Dina Belova
<br>
<br>
Software Engineer
<br>
<br>
Mirantis Inc.
<br>
<br>
</blockquote>
<br>
<blockquote type="cite">_______________________________________________
<br>
OpenStack-dev mailing list
<br>
<a class="moz-txt-link-abbreviated" href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a>
<br>
<a class="moz-txt-link-freetext" href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a>
<br>
</blockquote>
<br>
<br>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
OpenStack-dev mailing list
<a class="moz-txt-link-abbreviated" href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a>
<a class="moz-txt-link-freetext" href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a>
</pre>
</blockquote>
<br>
</body>
</html>