Thanks michoski, I have deployed CF2 to the 3rd data center, but I am having some issues hence these questions. I am still raw and learning CFEngine :-( Neil pointed me to his cfengine cookbook. Let me see if I could get my questions answered from there.
Thanks again. # mukarram On Thu, Feb 24, 2011 at 11:17 AM, michoski <micho...@cisco.com> wrote: > On 2/24/11 9:42 AM, "Mukarram Syed" <mukarram.s...@8x8.com> wrote: > > This is a good suggestion, but unfortunately I have inherited a CF2 > > environment and so I have to learn it. We will be moving to CF3 in the > near > > future. But the task in my hand is to migrate the CF2 infrastructure to > a new > > data center (the 3rd data center). We have CF2 running on 2 data > centers. > > Hence these questions. > > I have read through the Automating Unix and Linux Second Edition which > covers > > CF2 and couldn't find quite the answers I was looking for. I have also > > skimmed through the CF2 Reference Manual and the CF2 Tutorial as well. > > Many are and will be caught in this transitory phase for the next year or > so. The press and latest docs have all aligned with CF3, but even after > attending conferences and talking to others who have... CF2 will be used > into the foreseeable future in many large enterprises. It would be good if > a vendor could honestly admit this. It's not a CF problem mind you... > Cisco, for example, is the same way (use the latest, use the latest!). > > There used to be several sites I could point you to that cover exactly > this, > scaling cf2 to multiple data centers. There are different approaches you > can take, depending mostly on the number of clients and bandwidth. > > What we built was a "middle of the road" approach. First, we deployed one > large central cluster (several cfservd hosts behind a load balancer), which > also serves smaller sites over the WAN. The main site has around 1000 > managed nodes, with smaller sites hosting significantly less (100-200). > When we started building larger data centers similar in size to our first > site, we replicated the large load balanced cluster and used copy > statements > to slurp down workdir from the main VIP. This gave us local cfservds at > each of our larger sites, saving WAN bandwidth + latency. Within our > policies, we simply AllowRedefinitionOf things like policyhost, so each of > the larger sites point clients to the local VIP. > > Good luck! > >
_______________________________________________ Help-cfengine mailing list Help-cfengine@cfengine.org https://cfengine.org/mailman/listinfo/help-cfengine
