As I recall, the primary focus of CB_RECALL ANY was in dealing with recall of delegations. With delegations the server is in no position to, if needs to get some number back, pick ones that are not being used, since it is in the nature of delegations that the client may be using them with great benefit with the server not hearing anything about them. So the logic there was that if the client didn't do the picking and just did nothing the server would have to pick for himself, and would almost certainly do a poor job, and the client might not like the server's choices. For other recallable objects, the logic my be different. In the case of layouts, the data servers will know if they are being used but the MDFS will not. So if the client won't act on this, and the server picks badly, it can't say "there was no way I could have picked better ones" as it can with delegations. The spec should make it clear whether the client is at fault in this case or the server should have done its best by incorporating data server information on layout use, but .... The chances of client and server implementors agreeing on this are fairly low. So ambiguity/confusion on this issue may persist for a while. -----Original Message----- From: nfsv4-bounces@xxxxxxxx [mailto:nfsv4-bounces@xxxxxxxx] On Behalf Of Welch, Brent Sent: Monday, October 31, 2011 5:43 PM To: Jim Rees; Trond Myklebust Cc: linux-nfs@xxxxxxxxxxxxxxx; nfsv4 list Subject: Re: [nfsv4] [PATCH 2/2] nfs41: handle BLK_LAYOUT CB_RECALL_ANY CB_RECALL_ANY was added based on Panasas experience with its own protocols. When you have 100's (or 1000's) of clients and millions of files for which you maintain state, it is very much worth optimizing how the server can reclaim this state. Brent -----Original Message----- From: nfsv4-bounces@xxxxxxxx [mailto:nfsv4-bounces@xxxxxxxx] On Behalf Of Jim Rees Sent: Monday, October 31, 2011 11:32 AM To: Trond Myklebust Cc: linux-nfs@xxxxxxxxxxxxxxx; nfsv4 list Subject: Re: [nfsv4] [PATCH 2/2] nfs41: handle BLK_LAYOUT CB_RECALL_ANY Trond Myklebust wrote: I don't necessarily disagree with what you are saying, but I have yet to see a single server side implementation of CB_RECALL_ANY, let alone any numbers that indicate performance or responsiveness problems resulting from our existing client-side implementation. I therefore find it hard to understand why optimising this particular code is such a high priority, or why a patch that is adding per-file layoutreturns to initiate_bulk_draining() is going to help anything at all. Testing between the linux block layout client and the EMC block layout server revealed a deadlock when the server had handed out some number of layouts and couldn't hand out any more. So now the EMC block layout server implements CB_RECALL_ANY. So yes, this solves a real world problem, and yes, there is a server that implements this. We had some discussions at the time, and I don't remember if those were on the linux-nfs list or in some other forum. We decided that the client was in the best position to decide which layouts were no longer needed, so we needed some way for the server to tell the client to return some layouts without specifying which ones. CB_RECALL_ANY seemed custom-made for this purpose, so we used it. I don't think it would be appropriate for the server to recall all layouts when it only needs some of them back. _______________________________________________ nfsv4 mailing list nfsv4@xxxxxxxx https://www.ietf.org/mailman/listinfo/nfsv4 _______________________________________________ nfsv4 mailing list nfsv4@xxxxxxxx https://www.ietf.org/mailman/listinfo/nfsv4 -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
