環境
ESXi4.1
CentOS5.6 64bit
で動かしてるNFSがどうも遅い。
/var/log/messagesには以下のエラーが。
Apr 28 10:32:56 myhost kernel: INFO: task nfsd:4005 blocked for more than 120 seconds.
Apr 28 10:32:56 myhost kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Apr 28 10:32:56 myhost kernel: nfsd D ffff81003fe247e0 0 4005 1 4006 4004 (L-TLB)
Apr 28 10:32:56 myhost kernel: ffff81003acc5c00 0000000000000046 ffff81003acc5c10 ffffffff80062ff0
Apr 28 10:32:56 myhost kernel: ffff81003acc5b90 000000000000000a ffff81003de9b0c0 ffff81003fe247e0
Apr 28 10:32:56 myhost kernel: 000036f3fb0bf9d4 00000000000005d0 ffff81003de9b2a8 000000000004a0d1
Apr 28 10:32:56 myhost kernel: Call Trace:
Apr 28 10:32:56 myhost kernel: [<ffffffff80062ff0>] thread_return+0x62/0xfe
Apr 28 10:32:56 myhost kernel: [<ffffffff88036d8a>] :jbd:log_wait_commit+0xa3/0xf5
Apr 28 10:32:56 myhost kernel: [<ffffffff800a28b4>] autoremove_wake_function+0x0/0x2e
Apr 28 10:32:56 myhost kernel: [<ffffffff8803178a>] :jbd:journal_stop+0x1cf/0x1ff
Apr 28 10:32:56 myhost kernel: [<ffffffff8002fcd8>] __writeback_single_inode+0x1d9/0x318
Apr 28 10:32:56 myhost kernel: [<ffffffff800f5992>] sync_inode+0x24/0x33
Apr 28 10:32:56 myhost kernel: [<ffffffff8804c370>] :ext3:ext3_sync_file+0xcc/0xf8
Apr 28 10:32:56 myhost kernel: [<ffffffff884fc608>] :nfsd:nfsd_sync+0x7a/0xa4
Apr 28 10:32:56 myhost kernel: [<ffffffff884fcf70>] :nfsd:nfsd_commit+0x57/0x7a
Apr 28 10:32:56 myhost kernel: [<ffffffff8850347a>] :nfsd:nfsd3_proc_commit+0xbb/0xc8
Apr 28 10:32:56 myhost kernel: [<ffffffff884f91db>] :nfsd:nfsd_dispatch+0xd8/0x1d6
Apr 28 10:32:56 myhost kernel: [<ffffffff8848a649>] :sunrpc:svc_process+0x44c/0x713
Apr 28 10:32:56 myhost kernel: [<ffffffff80064604>] __down_read+0x12/0x92
Apr 28 10:32:56 myhost kernel: [<ffffffff884f95a1>] :nfsd:nfsd+0x0/0x2cb
Apr 28 10:32:56 myhost kernel: [<ffffffff884f9746>] :nfsd:nfsd+0x1a5/0x2cb
Apr 28 10:32:56 myhost kernel: [<ffffffff8005dfb1>] child_rip+0xa/0x11
Apr 28 10:32:56 myhost kernel: [<ffffffff884f95a1>] :nfsd:nfsd+0x0/0x2cb
Apr 28 10:32:56 myhost kernel: [<ffffffff884f95a1>] :nfsd:nfsd+0x0/0x2cb
Apr 28 10:32:56 myhost kernel: [<ffffffff8005dfa7>] child_rip+0x0/0x11
Apr 28 10:44:27 myhost kernel: mptscsih: ioc0: attempting task abort! (sc=ffff81003c7e9cc0)
Apr 28 10:44:27 myhost kernel: sd 0:0:1:0:
Apr 28 10:44:27 myhost kernel: command: Write(10): 2a 00 00 03 84 9f 00 00 08 00
Apr 28 10:44:27 myhost kernel: mptscsih: ioc0: task abort: SUCCESS (sc=ffff81003c7e9cc0)
調べると、死ぬほど長いディスカッションがされてた。
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=516374
http://linux.derkeiler.com/Mailing-Lists/Kernel/2009-01/msg10184.html
2.6系kernelのバグなのかな？
うーん。。。
exportsをasyncにして知らないふりをすることにした。
