Greetings folks,

With the migration of the Jenkins silo into the private cloud over the
weekend we wanted to make sure folks were aware of potential issues we
may see for a bit as we move forward.

* We are finding that between the production and sandbox silos we're
fairly regularly in an oversubscribed CPU state right now.

* As a result of the above it's very possible that jobs may be slower to
execute at present than previously

* Our ability to spin up as many systems as we did in the public cloud
is presently lower because of the above

Here's our current roadmap and planning on potential resolutions:

1) we're already actively working with Rackspace to do some
reconfiguration work in the public cloud to allow us to utilize it in a
manner more along how we've setup the private cloud

2) We're still actively working on tuning jobs to right size the amount
of CPU they actually need. Many of our build instances have historically
used more CPU than actually needed because Rackspace scales up the
number of CPU when you allocate an instance with more RAM. Since we have
complete control over the CPU / RAM allocation on instance definitions
in our private cloud we can tune things better

3) We will be discussing with Rackspace our needs for expanding the
hardware footprint needed to support the private cloud better.

Of these 1 and 2 will likely be our fastest to turn around. I don't have
an ETA on #1 just yet but as soon as some of the changes we need are
made Thanh and I will be focusing on testing the configuration updates
needed to support a hybrid cloud solution till #3 can come to fruition.

-Andy-

-- 
Andrew J Grimberg
Systems Administrator
Release Engineering Team Lead
The Linux Foundation

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Discuss mailing list
[email protected]
https://lists.opendaylight.org/mailman/listinfo/discuss

Reply via email to