> On Apr 26, 2016, at 10:13 AM, Steve Wise <swise@xxxxxxxxxxxxxxxxxxxxx> wrote: > > Hey Chuck, I'm testing this series on cxgb4. I'm running 'iozone -a -+d -I' on a share and watching the server stats. Are the starve numbers expected? Yes, unless you're seeing much higher numbers than you used to. > Every 5.0s: for s in /proc/sys/sunrpc/svc_rdma/rdma_* ; do echo -n "$(basename $s): "; cat $s; done Tue Apr 26 07:10:17 2016 > > rdma_stat_read: 379872 > rdma_stat_recv: 498144 > rdma_stat_rq_poll: 0 > rdma_stat_rq_prod: 0 > rdma_stat_rq_starve: 675564 This means work was enqueued on the svc_xprt, but by the time the upper layer invoked svc_rdma_recvfrom, the work was already handled by an earlier wake-up. I'm not exactly sure why this happens, but it seems to be normal (if suboptimal). > rdma_stat_sq_poll: 0 > rdma_stat_sq_prod: 0 > rdma_stat_sq_starve: 1748000 No SQ space to post a Send, so the caller is put to sleep. The server chronically underestimates the SQ depth, especially for FRWR. I haven't figured out a better way to estimate it. But it's generally harmless, as there is a mechanism to put callers to sleep until there is space on the SQ. > rdma_stat_write: 2805420 > > > On 4/25/2016 2:20 PM, Chuck Lever wrote: >> Second version of NFS/RDMA client patches proposed for merge into >> v4.7. Thanks in advance for any review comments! >> >> Attempt to fence memory regions after a signal interrupts a >> synchronous RPC. This prevents a server from writing a reply into a >> client's memory after the memory has been released due to a signal. >> >> Support providing a Read list and Reply chunk together in one RPC >> call. This is a pre-requisite for using krb5i or krb5p on RPC/RDMA. >> >> In addition, the following changes and fixes are included: >> >> - Use new ib_drain_qp() API >> - Advertise max size of NFSv4.1 callbacks on RPC/RDMA >> - Prevent overflowing the server's receive buffers >> - Send small NFS WRITEs inline rather than using a Read chunk >> - Detect connection loss sooner >> >> >> Available in the "nfs-rdma-for-4.7" topic branch of this git repo: >> >> git://git.linux-nfs.org/projects/cel/cel-2.6.git >> >> Or for browsing: >> >> http://git.linux-nfs.org/?p=cel/cel-2.6.git;a=log;h=refs/heads/nfs-rdma-for-4.7 >> >> >> Changes since v1: >> - Rebased on v4.6-rc5 >> - Updated patch description for "Avoid using Write list for ..." >> >> --- >> >> Chuck Lever (18): >> sunrpc: Advertise maximum backchannel payload size >> xprtrdma: Bound the inline threshold values >> xprtrdma: Limit number of RDMA segments in RPC-over-RDMA headers >> xprtrdma: Prevent inline overflow >> xprtrdma: Avoid using Write list for small NFS READ requests >> xprtrdma: Update comments in rpcrdma_marshal_req() >> xprtrdma: Allow Read list and Reply chunk simultaneously >> xprtrdma: Remove rpcrdma_create_chunks() >> xprtrdma: Use core ib_drain_qp() API >> xprtrdma: Rename rpcrdma_frwr::sg and sg_nents >> xprtrdma: Save I/O direction in struct rpcrdma_frwr >> xprtrdma: Reset MRs in frwr_op_unmap_sync() >> xprtrdma: Refactor the FRWR recovery worker >> xprtrdma: Move fr_xprt and fr_worker to struct rpcrdma_mw >> xprtrdma: Refactor __fmr_dma_unmap() >> xprtrdma: Add ro_unmap_safe memreg method >> xprtrdma: Remove ro_unmap() from all registration modes >> xprtrdma: Faster server reboot recovery >> >> >> fs/nfs/nfs4proc.c | 10 - >> include/linux/sunrpc/clnt.h | 1 >> include/linux/sunrpc/xprt.h | 1 >> include/linux/sunrpc/xprtrdma.h | 4 >> net/sunrpc/clnt.c | 17 + >> net/sunrpc/xprtrdma/backchannel.c | 16 + >> net/sunrpc/xprtrdma/fmr_ops.c | 134 +++++++-- >> net/sunrpc/xprtrdma/frwr_ops.c | 214 ++++++++------- >> net/sunrpc/xprtrdma/physical_ops.c | 39 ++- >> net/sunrpc/xprtrdma/rpc_rdma.c | 517 ++++++++++++++++++++++-------------- >> net/sunrpc/xprtrdma/transport.c | 16 + >> net/sunrpc/xprtrdma/verbs.c | 91 ++---- >> net/sunrpc/xprtrdma/xprt_rdma.h | 42 ++- >> net/sunrpc/xprtsock.c | 6 >> 14 files changed, 674 insertions(+), 434 deletions(-) >> >> -- >> Chuck Lever >> -- >> 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 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-rdma" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- Chuck Lever -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
