Description of problem: Currently there are two variations of readdirp: 1. Directories are read from all the bricks and the client filters the duplicate directory entries, by removing the entry if it doesn't hash to the brick that returned the readdirp. 2. dht-readdir-optimise option on: In this case, the directory is read from only one brick(first up subvol) and dht indicates all other bricks to filter the directory entries(at the posix layer) 1- is less performant, while 2 has the problem that some directories can go unlisted if they are not healed to the first up subvol. Hence to get the performance of option 2 and reliability of option 1: we can have a dht stub on the server side which filters the directories in readdirp if they do not hash to that brick. This was proposed by Rghavendra Gowdappa Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
REVIEW: http://review.gluster.org/16312 (dht-server-stub: Add a server side dht to filter the unhashed dir in readdirp) posted (#5) for review on master by Poornima G (pgurusid)
REVIEW: https://review.gluster.org/16824 (dht: move hash calculation code to libglusterfs) posted (#1) for review on release-3.10 by Poornima G (pgurusid)
REVIEW: https://review.gluster.org/16825 (dht-server-stub: Add a server side dht to filter the unhashed dir in readdirp) posted (#1) for review on release-3.10 by Poornima G (pgurusid)
I'm closing this BZ with WontFix. Please reopen the BZ if you feel this should be fixed.