[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] machine for central-manager
- Date: Thu, 22 Apr 2010 10:11:30 +0100
- From: Santanu Das <santanu@xxxxxxxxxxxxxxxxx>
- Subject: Re: [Condor-users] machine for central-manager
Many thanks Ian, James and Dan for the useful info and the link. It
really helps a lot to get things started. I'm gonna run collector +
negotiator + view_server on that node and I'm gonna use a Dell PE R410,
which is 8-core (2*quad-core E5540 each) with 16G of memory. All my
nodes are EGEE (formally gLite and soon-to-be EGI) Worker nodes and I
plan to host site_BDII + MON + Condor Central manager as three virtual
guests off that machine. According to your experiences with VM, I think
the proposed ESXi node is more than adequate for the job. I'm gonna
give it a try shortly and report here back as it goes.
Thanks again for your time.
James Osborne wrote:
We have run our central managers on ESXi (3.5, soon moving to 4.0 or
vSphere) for over 18 months now managing a pool of Windows XP machines
spread across our campus providing access to almost 6000 cores. It
runs on a fraction of 1 core of an 8 core ESX box and has 2GB of RAM
allocated to it. The ESXi box is an 8 core system containing two
Intel E5472 processors running at 3.0GHz with a total of 32GB of RAM.
The ESXi box is connected to a resilient NFS backing store hosting
the virtual machine files.
Over the last 6 months the series 7 central manager has maxed out at
15% of 1 core and runs at an average of 7%, maximum memory of 80% of
2GB and runs at an average of 38%, maximum network IO of 101 Kbps and
an average of 35 Kbps. We have also virtualized submit machines as
well, so an ESX box like ours with a good internal RAID would be more
than sufficient as a head node providing services around a 400-core
cluster using Condor.
You should probably talk to your local HPC people before going ahead
with any development and seek their advice directly
> Dear all,
> How powerful a machine (CPU-core, memory etc.) should be to run condor
> central-manager for a ~400-core cluster? I'm interested in
> so I wanna get a rough idea about the hardware we should go for to
> accommodate all the services/virtual machines to run nicely.
> Also, is there any recommendation on vitalization platform as far as
> condor is concerned? Is VMware ESXi a good choice??
> Thanks in advance for any help.
> Condor-users mailing list
> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx
> subject: Unsubscribe
> You can also unsubscribe by visiting
> The archives can be found at: