On 07/05/2009 08:05 PM, Shawn Wells wrote:
On 07/05/2009 03:59 AM, Andrew Avramenko wrote:
Shawn,
Have Red Hat any plans to extend kernel with features which are
already in SLES11 such as dynamic extending of memory in the RHEL 5
branch or we should wait for RHEL 6 for that?
Certain ones, 100% based upon customer (or potential customer) demand.
We've had a few customers request dynamic memory to be backported into
RHEL5, and we're tracking the request through BugZilla 507705
[https://bugzilla.redhat.com/show_bug.cgi?id=507705]. The ETA on this
would be RHEL 5.5. The process is fairly trivial, as the code is
already upstream in the 2.6.27 kernel.
Generally speaking, it takes us a minor release (i.e. RHEL 5.4 to RHEL
5.5) to backport a feature that's already upstream. Both the Red Hat &
IBM engineering team has to backport & test the patch, then agree to
support it, which takes 2-3mo. Customer requests make this process go
smoothly... so if there is something out there people are interested in,
let me know.
If there is a feature list people are interested in, feel free to reply
on/off list.
I cleared out my EMail folders over the weekend, so I'm going to thread
jack this to answer some of the other questions....
from John:
I have a Q on the RHEL5 list about access for on-customers (just posted
a few minutes ago).
You can get a free 180 day eval (which includes access to betas) at
http://www.redhat.com/z. On the left side you'll see a link called
"Free Evaluation." Once you download the bits, simply use the
activation key that was sent out in the beta announcement. To save you
some hunting for it: 49af-8941-4d14-7589
from Scott:
Hmmm.. Disabled cpu topology, added processor degradation capability...
Those sound, well.. bad.
The devil is in the details. Here's the release notes for those
features from BugZilla... they're ultimately good things.
<snip>
Problem: If the hardware supports the configuration-topology facility
the system cpu topology information will be passed to the
scheduler. The scheduler bases its load balancing decisions
also on the cpu topology. This may have a negative performance
impact on machines where I/O interrupts are distributed unevenly
to cpus. Those cpus which are grouped together and get more I/O
interrupts than others will tend to have a higher average load.
Solution: Disable cpu topology support by default and add a kernel parameter
which allows to switch it on. In order to switch support on
topology=on needs to be added to the kernel parameter line.
</snip>
And regarding processor degradation:
<snip>
z990, z890 and later systems may reduce processor speed in case of cooling
problems. This is observed thru SCLP event type 4 event qualifier 3 (processor
capacity change).
Further, STSI will report the new capacity. This feature triggers an uevent and
issues issue a
warning to Syslog. This enables automation software to observe the machine
state and act based on
policies, as applicable.
</snip>
lastly, from Stephen:
Shawn could you provide a url/link for downloading this?
System z:
https://rhn.redhat.com/rhn/software/channel/downloads/Download.do?cid=6019
x86:
https://rhn.redhat.com/rhn/software/channel/downloads/Download.do?cid=6002
x64_64:
https://rhn.redhat.com/rhn/software/channel/downloads/Download.do?cid=6007
Itanium:
https://rhn.redhat.com/rhn/software/channel/downloads/Download.do?cid=6012
PPC:
https://rhn.redhat.com/rhn/software/channel/downloads/Download.do?cid=6017
If you don't have a RHN login, refer back up to Johns' question.
Cheers,
Shawn
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
