I haven't used Puppet but you should also be able to write your own small script awk/sed or some shell commands to format that is required by Puppet. On Mon, May 9, 2011 at 1:03 AM, Horacio Sanson <hsanson at gmail.com> wrote: > On Mon, May 9, 2011 at 9:59 AM, Anand Babu Periasamy <ab at gluster.com> wrote: >> On Mon, May 9, 2011 at 5:37 AM, Horacio Sanson <hsanson at gmail.com> wrote: >>> >>> Where can I find documentation about manual configuration of Gluster >>> peers/volumes? All documentation seems to be about the gluster CLI. I >>> would prefer manual configuration to facilitate automation via scripts >>> (e.g. Puppet/Chef). >>> >>> I also read in this list that it is possible to configure Raid10 via >>> text files... ?I would also like to experiment with this setup. Any >>> related documents on how to do this would be appreciated. >>> >>> regards >>> Horacio >> >> >> Hi Horacio, >> CLI also coordinates with elastic volume manager. You will lose the ability >> to perform online volume management operations. Another main reason for >> abstraction is to avoid untested combinations. It is possible to use the >> auto-generated volume spec files (from /etc/gluster dir) and start >> glusterfsd manually. My strong recommendation is to integrate pupper/chef >> with Gluster CLI. Integration with CLI takes fewer lines of code than manual >> vol spec approach. What trouble did you run in to? ?I think it makes sense >> for us to document Gluster / Puppet integration and provide examples. > > One of the problems I have with gluster/puppet integration is the > output format of gluster commands is not parse friendly. In puppet I > need to add onlyif and unless options so the gluster commands don't > get executed every time. In the case of trusted pools the peer command > I can easily check if a peer belongs to a pool: > > ? gluster peer status | grep -q -s 192.168.4.91 ? ?# Returns 0 if the > peer exists or 1 if not > > but is more complicated to check if a particular peer is connected, > disconnected, rejected or whatever status since this value is in a > separate line. If the IP address and status were in a single line all > would be very easy. > > >> RAID10 like is simply distributed volume with replica count = 2. >> -- >> Anand Babu Periasamy >> Blog [http://www.unlocksmith.org] >> >> Imagination is more important than knowledge --Albert Einstein >> > > regards, > Horacio > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users >
