From mboxrd@z Thu Jan 1 00:00:00 1970 From: Benny Halevy Subject: Re: pNFS timeouts Date: Mon, 14 Jun 2010 10:11:30 -0400 Message-ID: <4C163892.20201@panasas.com> References: <20100614085157.GS17703@ics.muni.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-2 Cc: "J. Bruce Fields" , linux-nfs@vger.kernel.org To: Lukas Hejtmanek Return-path: Received: from daytona.panasas.com ([67.152.220.89]:37435 "EHLO daytona.int.panasas.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1753276Ab0FNOLa (ORCPT ); Mon, 14 Jun 2010 10:11:30 -0400 In-Reply-To: <20100614085157.GS17703@ics.muni.cz> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Jun. 14, 2010, 4:51 -0400, Lukas Hejtmanek wrote: > Hi, > > I'm playing with pNFS (spnfs layouts) using one MDS and one DS. I have > a client that mount pNFS volume. > > I can touch (create) files on that volume immediately: > time touch c > > real 0m0.005s > user 0m0.000s > sys 0m0.000s > > But creating files with a content suffers from approx. 15 secs timeout: > > dd if=/dev/zero of=testf10 bs=1M count=1 > 1+0 records in > 1+0 records out > 1048576 bytes (1.0 MB) copied, 15.1156 s, 69.4 kB/s > > overwriting the same file works OK: > > dd if=/dev/zero of=testf10 bs=1M count=1 > 1+0 records in > 1+0 records out > 1048576 bytes (1.0 MB) copied, 0.0258294 s, 40.6 MB/s > > I think this is an interesting part of the debug logs: > [69615.791232] pnfs_writeback_done: Begin (status -10008) I'm not sure if that explains the 15 seconds delay. I wonder if it could be related to the nfsv4 grace period. Does this happen if you wait for a couple minutes after the server restarts before starting the test? Benny > [69615.791238] put_lseg: lseg ffff88003b987840 ref 7 valid 1 > [69615.794450] NFS: 6195 nfs_writeback_done (status -10008 count 8192) > [69615.794450] pnfs4_write_done DS write > [69615.794450] nfs41_sequence_done: Error 0 free the slot > [69615.794450] nfs4_free_slot: free_slotid 0 highest_used_slotid 5 > [69615.794450] <-- pnfs4_write_done status= -11 > [69615.990137] filelayout_write_call_done new off 8192 orig offset 8192 > > Any ideas? >