Hi, I left a nfsometer running last weekend. Today I tried to get the results but for my sadness the nfsometer can't complete the tests. It stopped on filebench_fileserver workload. The dmesg has some interesting messages like about the filebench tasking hanging for more than 120 seconds. And if I try to run it again I get "NFS client not idle" error. How can I handle this error? I don't want to reboot before understanting what is happening. There is some key points here. The server is a freebsd11 and the client a centos7 (kernel 3.10). Also the server is besides an infiniband card by using ipoib. The MTU is setted to 65K because of ipoib. [ 4800.645708] INFO: task filebench:4790 blocked for more than 120 seconds. [ 4800.645758] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 4800.645816] filebench D ffff8820214e5870 0 4790 4756 0x00000080 [ 4800.645819] ffff882030b2bc70 0000000000000082 ffff88203da40fb0 ffff882030b2bfd8 [ 4800.645823] ffff882030b2bfd8 ffff882030b2bfd8 ffff88203da40fb0 ffff8820214e5868 [ 4800.645826] ffff8820214e586c ffff88203da40fb0 00000000ffffffff ffff8820214e5870 [ 4800.645830] Call Trace: [ 4800.645835] [<ffffffff8168cca9>] schedule_preempt_disabled+0x29/0x70 [ 4800.645839] [<ffffffff8168a905>] __mutex_lock_slowpath+0xc5/0x1c0 [ 4800.645844] [<ffffffff81689d6f>] mutex_lock+0x1f/0x2f [ 4800.645847] [<ffffffff8120cd9f>] do_last+0x28f/0x12a0 [ 4800.645852] [<ffffffff811de4f6>] ? kmem_cache_alloc_trace+0x1d6/0x200 [ 4800.645856] [<ffffffff8120de72>] path_openat+0xc2/0x490 [ 4800.645866] [<ffffffffa09ac60d>] ? nfs_attribute_cache_expired+0x2d/0x70 [nfs] [ 4800.645869] [<ffffffff8120ffeb>] do_filp_open+0x4b/0xb0 [ 4800.645873] [<ffffffff8121cc67>] ? __alloc_fd+0xa7/0x130 [ 4800.645878] [<ffffffff811fd4f3>] do_sys_open+0xf3/0x1f0 [ 4800.645882] [<ffffffff811fd60e>] SyS_open+0x1e/0x20 [ 4800.645887] [<ffffffff81696b09>] system_call_fastpath+0x16/0x1b [root@centos7 ~]# nfsometer 1.2.3.4:/tank/nfsshare filebench_fileserver > No options specified. Using default: -o v3 -o v4 -o v4.1 > Using results directory: /root/nfsometer_results NFS client not idle (check /proc/fs/nfsfs/servers) run "nfsometer --help" and "nfsometer examples" for more info [root@centos7 ~]# Best regards, -- 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
