[Gluster-devel] Two questions about the performance of ls in glusterfs

[Keiviw]

hi,
I have two questions about the performance of ls in glusterfs,and I know that  the community has noticed the poor performance of ls and made something to solve the problem,like readdir-ahead.Here are my questions.
1,The preload  (a readdir request not initiated by application, but instead triggered by readdir-ahead in an attempt to pre-emptively fill the read-ahead buffer) is in process,a readdir request from application waits for its completion. In the code, when the preload is in progress,it locks the ctx. For applications, the readdir-ahead's request processing is synchronous, is it possible that the preload handles request asynchronously by reducing lock granularity? In other words, the client fetches the dentries from the preload,while the preload gets the dentries from the servers.
2, when does it can be a lager buffer,like io-cahce,which cached the data or dentries read before? As you know, ls is so slow.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160229/c8f84ada/attachment.html>