[Spacewalk-list] How to improve WUI performance

Thu Oct 13 19:19:04 UTC 2016

Jan-Albert,

Thank you for the information, this and your previous reply.  Both very helpful.

Daryl

________________________________
From: spacewalk-list-bounces at redhat.com <spacewalk-list-bounces at redhat.com> on behalf of Ree, Jan-Albert van <J.A.v.Ree at marin.nl>
Sent: Thursday, October 13, 2016 12:41 PM
To: spacewalk-list at redhat.com
Subject: Re: [Spacewalk-list] How to improve WUI performance

Pushing larger updates to ALL clients at once is a bad idea as soon as you go over 20-30 clients already...

Our SW server (Scientific Linux 7, SW 2.5) is physical, with 64GB RAM. 6 cores and a solid RAID drive set. As soon as I for instance update some software on one of our HPC clusters, as soon as I allow more than 20-25 clients to pull simultaneous I start getting tracebacks in my mailbox... specially if the updates become larger (new kernels, new nVidia CUDA suite, new LibreOffice)  I would try to move to CentOS/RHEL/ScientificLinux 7 for your new install, the later PostgreSQL version in it is a lot more efficient which helps speed up your SW instance.

When I moved to a new server I didn't want to pull all the history (and some poor choices/mistakes from the past) so I set up a clean new server, with the proper channels/configs/packages and then on the old server selected batches of clients to which I pushed a script which registered them with the new server.

This worked very well for us and allowed us to test and tweak the setup (it was during this phase on the new server that we noticed the OSAD issues and managed to solve it with moving to sqlite3) before moving to full production.

What helped us debug/fix OSA issues was the script from post https://www.redhat.com/archives/spacewalk-list/2014-May/msg00124.html ; great to easily track which nodes are not working as expected.

Regards,

--

Jan-Albert

Jan-Albert van Ree | Linux System Administrator | MARIN Support Group
MARIN | T +31 317 49 35 48 | J.A.v.Ree at marin.nl<mailto:J.A.v.Ree at marin.nl> | www.marin.nl<http://www.marin.nl>

[LinkedIn]<https://www.linkedin.com/company/marin> [YouTube] <http://www.youtube.com/marinmultimedia>  [Twitter] <https://twitter.com/MARIN_nieuws>  [Facebook] <https://www.facebook.com/marin.wageningen>
MARIN news: Software seminar in Shanghai for the first time<http://www.marin.nl/web/News/News-items/Software-seminar-in-Shanghai-for-the-first-time.htm>

________________________________
From: spacewalk-list-bounces at redhat.com <spacewalk-list-bounces at redhat.com> on behalf of Konstantin Raskoshnyi <konrasko at gmail.com>
Sent: Thursday, October 13, 2016 19:19
To: spacewalk-list at redhat.com
Subject: Re: [Spacewalk-list] How to improve WUI performance

Hm..do you use osad? Just try to push updates for all of your clients and see what happens ;)

On Thu, Oct 13, 2016 at 10:08 AM, Coffman, Anthony J <Tony.Coffman at snapon.com<mailto:Tony.Coffman at snapon.com>> wrote:

Spacewalk 2.5 (with 3 remote proxies)
CentOS 6.8 VM (running stock Postgres 8.4)
4 vCPUs
12GB memory

550 clients (CentOS/OracleLinux/EL5/6/7 with a sprinkling of Ubuntu and SuSE thrown in)A

The System page in the WebUI loads in about 4-6 seconds (faster when the cache is warm).

I’ve done essentially no tuning except for changing the proxy timeout (due to issue with the large updateinfo size for Oracle Linux).

Looking at our memory utilization we could probably get away with 6-8GB of memory for our workload.

Hope this data point is useful.

Regards,
--Tony

From: spacewalk-list-bounces at redhat.com<mailto:spacewalk-list-bounces at redhat.com> [mailto:spacewalk-list-bounces at redhat.com<mailto:spacewalk-list-bounces at redhat.com>] On Behalf Of Daryl Rose
Sent: Thursday, October 13, 2016 12:25 PM
To: spacewalk-list at redhat.com<mailto:spacewalk-list at redhat.com>
Subject: Re: [Spacewalk-list] How to improve WUI performance

I am actually leaning to moving to a physical server.  I have a server in the closet that is dual CPU, 8 core, 64GB of RAM.  I was originally thinking that it was a bit of over kill, but after reading this reply, perhaps its sized correctly.

If I do move to a physical, can I migrate everything from the original SW server to the replacement server?  Can I export the database and import it over?  Migrate repositories over?  I am using a signed certificates, and I would like to continue using the same machine name and configuration.  Do I have to re-register these servers?  Its taken me a while to get 465 added in, and I don't want to have to re-register everything, especially since I continue adding on a daily bases.

Thanks

Daryl

________________________________
From: spacewalk-list-bounces at redhat.com<mailto:spacewalk-list-bounces at redhat.com> <spacewalk-list-bounces at redhat.com<mailto:spacewalk-list-bounces at redhat.com>> on behalf of Konstantin Raskoshnyi <konrasko at gmail.com<mailto:konrasko at gmail.com>>
Sent: Thursday, October 13, 2016 10:43 AM
To: spacewalk-list at redhat.com<mailto:spacewalk-list at redhat.com>
Subject: Re: [Spacewalk-list] How to improve WUI performance

You need bare metal high performance machine.
We have 400 machines and use xeon7 24cores with 64 of ram.
Plus tuning: Postgres, Linux, tomcat

On Thursday, October 13, 2016, Daryl Rose <darylrose at outlook.com<mailto:darylrose at outlook.com>> wrote:

I have 465 servers in my SW environment, and I keep adding more everyday.  When finished, I'll have nearly a thousand servers in the environment.

My SW server is a virtual RHEL 6.6, SW v2.3.  I have four CPU's and four Gigs of memory allocated to the server.

Whenever I select "Systems" from the main menu a single CPU pegs at 100%, and it takes several minutes for the page to come up.  It appears that osa-dispatcher is what is using up the CPU.  osa-dispatcher is often at 99 to 100%.

Is there  a way to performance tune SW?  Can I configure SW to use all four CPU's, or should I add additional CPU's, can configure SW to use all assigned CPU's?

Thanks

Daryl

_______________________________________________
Spacewalk-list mailing list
Spacewalk-list at redhat.com<mailto:Spacewalk-list at redhat.com>
https://www.redhat.com/mailman/listinfo/spacewalk-list<https://portal.marin.nl/mailman/listinfo/,DanaInfo=www.redhat.com,SSL+spacewalk-list>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20161013/204b672c/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: imageabcd9b.PNG
Type: image/png
Size: 293 bytes
Desc: imageabcd9b.PNG
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20161013/204b672c/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: imagec70e4d.PNG
Type: image/png
Size: 331 bytes
Desc: imagec70e4d.PNG
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20161013/204b672c/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: imageacccc5.PNG
Type: image/png
Size: 333 bytes
Desc: imageacccc5.PNG
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20161013/204b672c/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image7fd843.PNG
Type: image/png
Size: 253 bytes
Desc: image7fd843.PNG
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20161013/204b672c/attachment-0003.png>