I'm sure my RH experience is applicable to CentOS as well.

1. We've Cfengine it in limited production for a while on RH9 and RHEL3 workstations. 2. Don't yet have ant RHEL4 workstations in production, and we haven't yet implemented it for servers other than running a Cfengine server. 3. Cfengine is an extremely complex package, and the learning curve is quite steep. Map out what you want to do carefully. and have another associate eyeball your design. Standard practice. 4. I've done very little with "pushing" rpms to the workstations, but I'm now in the process of preparing such an upgrade package. 5. We have now implemented a test Cfengine server (perhaps even more than one) and workstations to avoid the "oops, I thought that would work" conditions. 6. Probably a good idea to keep the Cfengine inputs under CVS/SVN control. We're thinking about it. 7. Treat Cfengine system like production and be careful with changes. See #5 above. 8. We didn't install from RPM. I'm sure that would be fine, but install changes iin test first. Standard practice.

HTH,

-- Collins Richey If you fill your heart with regrets of yesterday and the worries of tomorrow, you have no today to be thankful for.