Ok, will make the changes and send it as a single tcp_nodelay patch. One final clarification, do we want the TCP_NODELAY set by default or only when requested?
-----Original Message----- From: Ilya Dryomov [mailto:[email protected]] Sent: Thursday, January 22, 2015 8:38 PM To: Chaitanya Huilgol Cc: Somnath Roy; Sage Weil; Ceph Development Subject: Re: [PATCH] ceph: rbd option listing and tcp_nodelay support On Thu, Jan 22, 2015 at 5:39 PM, Chaitanya Huilgol <[email protected]> wrote: > Well, there was no ceph_options pointer in the messenger layer and no_crc > flag was being passed into the messenger as an additional flag and maintained > within the messenger again as bool no_crc, I inferred that it was intentional > and we did not want the generic options in the messenger layer. If it is > fine to have messenger point back to the ceph_options, then that’s the > easiest way to do it. > Also, we would have some additional option coming in with RDMA support (xio > messenger) which would be very messenger specific, so having a messenger > options had made sense. OK, we'll see how to refactor it nicely then. Given that tcp_nodelay is just a bool, for now I'd suggest the simplest and least invasive patch: - add bool tcp_nodelay field to ceph_messenger - add bool tcp_nodelay parameter to ceph_messenger_init() - call ceph_messenger_init() with ceph_test_opt(client, TCP_NODELAY) Thanks, Ilya ________________________________ PLEASE NOTE: The information contained in this electronic mail message is intended only for the use of the designated recipient(s) named above. If the reader of this message is not the intended recipient, you are hereby notified that you have received this message in error and that any review, dissemination, distribution, or copying of this message is strictly prohibited. If you have received this communication in error, please notify the sender by telephone or e-mail (as shown above) immediately and destroy any and all copies of this message in your possession (whether hard copies or electronically stored copies).
