scaling cfengine?

[Tom Perrine]

-----BEGIN PGP SIGNED MESSAGE-----

We've been running cfengine for about 5 years, and we're up to about
320 machines running on Solaris, AIX, Linux, Tru64, IRIX, etc.

These are mostly using cfengine 1.x with 2.x being phased in.  We
don't use the daemon stuff; we mount the cfengine data area read-only
over NFS and "pull" the updates.

We have not seen any scaling problems yet, but we're thinking of
managing some 1000 node clusters with cfengine in the next year or so.

In our current setup, every machine runs cfengine at boot time, and
every morning.  So far, all we've had to do is stagger the morning
cfengine runs to spread the network and fileserver load across several
hours, which is a holdover from our 10Mbit networking days.

Has anyone seen *any* scaling problems with cfengine at "larger"
sites?  How is performance of the daemon these days?  What's the
largest number of hosts that people are managing as a single site?

I'm a *very* happy member of the cult of cfengine, I can't imagine
running any site with more than 3 computers without it.  But I need
to convince some cluster people NOT to go write their own software
update system.

- -- 
Tom E. Perrine <tep@SDSC.EDU> | San Diego Supercomputer Center 
http://www.sdsc.edu/~tep/     | 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.7 <http://mailcrypt.sourceforge.net/>

iQCVAwUBPbbccxTSxpWcaAFRAQGNoQQAj6jHU3VCTj7pgpN1zOsJfxpkhIkLr5YH
qgp6n5iD79K1CSwXeZ/pHzeIFJywoxcIu/RRhDAhCQo8TkF/rffKWW3rfwoTkoOS
qMGV+m5RkfoQnxGi2aLxI3H9goyvziU4E/oGx5D7DMZew4Cst7sBObDq6hpajrMz
G+yNjwHepnM=
=uRMF
-----END PGP SIGNATURE-----