Hi, the following few patches add the `FETCH' command to rrdcached, adds a client interface and changes `rrdtool fetch' and `rrdtool graph' to receive values from the daemon directly if they are connected to one.
This allows to create graphs on machines that don't have direct access to the RRD files. It is common to keep the collection and graphing part on separate machines, usually for performance and security reasons. With these patches, setting up such a system becomes a lot easier because no shared storage is needed anymore. The first two patches are pretty straight forward, though maybe not very nice to read. The third and fourth patches are mostly meant for discussion. I'd be interested to know whether you'd prefer this approach (use `FETCH' if connected to the daemon) or a similar approach to that used by Scott Brumbaugh back in May 2008: Specify the daemon address as part of the filename. The second approach would have the advantage, that data from more than one daemon could be used in a single graph (yes, I hear people already are in that situation). The disadvantage is that handling `graph's and `fetch'es may alter your established connection. To solve the latter, I think the best solution would be to pass a `connection object' to each `rrdc_*' function as first argument. Any feedback welcome :) Regards, -octo _______________________________________________ rrd-developers mailing list rrd-developers@lists.oetiker.ch https://lists.oetiker.ch/cgi-bin/listinfo/rrd-developers
