[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: openshift memory requirements



----- Original Message -----
> From: "Hetz Ben Hamo" <hetz hetz biz>
> To: "Clayton Coleman" <ccoleman redhat com>
> Cc: "users" <users lists openshift redhat com>
> Sent: Tuesday, 23 May, 2017 6:12:16 PM
> Subject: Re: openshift memory requirements
> 
> ouch, is there another solution? I just want to show some pretty graphs for
> cpu/network/memory, not to collect them or anything...

If you are running this on a small cluster, you can set a memory limit and the metric pods will configure themselves to that. You might not get the best performance out of it and this may not work if you are dealing with a larger cluster size, but is more than possible to restrict the memory size to something smaller. I believe 'oc cluster up' is already using more restricted values for the metrics memory.

To start I would configure the Hawkular Metrics and Cassandra to use something like 512 MB each and see if that works for you. You can then adjust from there.

> 
> תודה,
> חץ בן חמו
> אתם מוזמנים לבקר בבלוג היעוץ או בבלוג הפרטי שלי
> 
> On Wed, May 24, 2017 at 1:03 AM, Clayton Coleman < ccoleman redhat com >
> wrote:
> 
> 
> 
> Metrics requires about 4gb by itself
> 
> On May 23, 2017, at 5:51 PM, Louis Santillan < lsantill redhat com > wrote:
> 
> 
> 
> 
> If the machine is an i7 it likely only has 4 cores/threads total. 3 VM x 2
> cores = 6 cores required. Also, instead of having 3 VMs with at least one IO
> controller and a NIC each, you have 3 VMs sharing 1 IO controller and 1 NIC.
> Not as fun as it sounds.
> 
> My lab machine is a HP z820 with 32 cores and 128GB of RAM. When I spin up a
> SmartStart style 3x3x3 cluster + 1 bastion host + 1 NFS host (3 masters, 3
> infra, 3 app nodes + 2 support hosts), my IO controller can't keep up. I
> might as well be running on some 4 year old cellphone. Which reminds me, I
> need to move those VMs over to the storage attached to my RAID controller.
> 
> 
> 
> -----------------------------------------------------------------------
> 
> LOUIS P. SANTILLAN
> 
> SENIOR CONSULTANT, OPENSHIFT, MIDDLEWARE & DEVOPS
> 
> Red Hat Consulting, NA US WEST
> 
> lpsantil gmail com M: 3236334854
> TRIED. TESTED. TRUSTED.
> 
> On Tue, May 23, 2017 at 11:13 AM, Hetz Ben Hamo < hetz hetz biz > wrote:
> 
> 
> 
> Well, I installed (through the Ansible install) the origin 1.5 on 3 VM's,
> each of them had 4 GB RAM, the nodes had 2 cores.I also enabled the metrics.
> 
> The entire system barely responded (this is an i7 machine with 16GB RAM).
> Thats why I asked if there are any changes that I need to add through
> Ansible to make the system work well with such amount of RAM.
> 
> תודה,
> חץ בן חמו
> אתם מוזמנים לבקר בבלוג היעוץ או בבלוג הפרטי שלי
> 
> On Tue, May 23, 2017 at 9:09 PM, Clayton Coleman < ccoleman redhat com >
> wrote:
> 
> 
> 
> OpenShift at that scale probably requires 300-500M on the master and 100-200M
> at the nodes at that scale.
> 
> On Tue, May 23, 2017 at 2:02 PM, Hetz Ben Hamo < hetz hetz biz > wrote:
> 
> 
> 
> I thought about 2 nodes with 4GB (and 2 coers) at the beginning and add nodes
> as needed.
> Number of pods - thought about using 6 (2 wordpress, 2 mysql, 2 memcache)
> 
> תודה,
> חץ בן חמו
> אתם מוזמנים לבקר בבלוג היעוץ או בבלוג הפרטי שלי
> 
> On Tue, May 23, 2017 at 8:59 PM, Clayton Coleman < ccoleman redhat com >
> wrote:
> 
> 
> 
> How many nodes and pods are you planning to run?
> 
> On Tue, May 23, 2017 at 1:43 PM, Hetz Ben Hamo < hetz hetz biz > wrote:
> 
> 
> 
> Hi,
> 
> I've read the docs about openshift memory requirements and I wanted to ask
> something..
> 
> I'm planning to build a system which will host a web site (wordpress based,
> for example) which will auto-scale based on the number of visitors..
> 
> According to the docs in the link (
> https://docs.openshift.com/container-platform/3.5/install_config/install/prerequisites.html
> ) it requires 16GB RAM (it's used to be 8GB in 3.0). It doesn't mention the
> amount of ram needed for the infra nodes in the docs (I assume another 8GB?)
> 
> My question: is there any way to build a system with much less ram? something
> like 4GB for Master, 4GB per node (minimum 2 nodes)? if so, what
> configurations should I add to my ansible host file?
> 
> Thanks
> 
> _______________________________________________
> users mailing list
> users lists openshift redhat com
> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
> 
> 
> 
> 
> 
> 
> _______________________________________________
> users mailing list
> users lists openshift redhat com
> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
> 
> 
> 
> 
> _______________________________________________
> users mailing list
> users lists openshift redhat com
> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
> 


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]