Description of problem: By default NFS-Ganesha uses it own caching layer(MDCACHE), nfs-ganesha is not getting improvements offered by gluster md cache. There is no point in having enabled both together. This bug is validate performance mdcache with ganesha in which we disable Ganesha mdcache using following parameters(may be more) CACHEINODE { Entries_HWMark = 1; Attr_Expiration_Time = 0; }
Converting this to upstream for the time being and if we are confident enough I will open a new bug in downstream
Migrated to github: https://github.com/gluster/glusterfs/issues/606 Please follow the github issue for further updates on this bug.