In my memory the reason the files are written to disk is because nobody wrote a input/output stream or other io interfaces for the XML parser to use. Easy enough to do through. On Fri, Nov 18, 2016 at 3:50 PM Jinmei Liao <jil...@pivotal.io> wrote:
> As far as I remember, cluster configurations files are saved in an internal > region as well, and that region is defined as replicate_persistent, so the > content of it gets written to the disk as well in those ConfigDistDir > backup files. This is something we need to clean up as well. Just use one > way of saving the content, either in file system or internal region. I > would prefer internal region since it will do data propagation and > persistence for us for free, but we still need to have a place to save the > jars as well. > > On Sat, Nov 19, 2016 at 12:32 AM, Jared Stewart <jstew...@pivotal.io> > wrote: > > > The cluster.xml and cluster.properties files seem to contain the entirety > > of a given cluster configuration. These are the only files included in > the > > zip file produced by ‘export cluster-configuration’. Does anyone know > what > > the ConfigDiskDir files are used for? Are they just redundant versions > of > > the information stored in the .xml and .properties files? > > > > ./cluster_config > > ./cluster_config/cluster > > ./cluster_config/cluster/cluster.properties > > ./cluster_config/cluster/cluster.xml > > ./ConfigDiskDir_l1 > > ./ConfigDiskDir_l1/BACKUPcluster_config.if > > ./ConfigDiskDir_l1/BACKUPcluster_config_1.crf > > ./ConfigDiskDir_l1/BACKUPcluster_config_1.drf > > ./ConfigDiskDir_l1/DRLK_IFcluster_config.lk > > > > > -- > Cheers > > Jinmei >
