From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx3.redhat.com (mx3.redhat.com [172.16.48.32]) by int-mx1.corp.redhat.com (8.11.6/8.11.6) with ESMTP id jA127aV01474 for ; Mon, 31 Oct 2005 21:07:36 -0500 Received: from ms1.nttdata.co.jp (ms1.nttdata.co.jp [163.135.193.232]) by mx3.redhat.com (8.13.1/8.13.1) with ESMTP id jA127RPL009482 for ; Mon, 31 Oct 2005 21:07:28 -0500 Received: from ms1.nttdata.co.jp (localhost [127.0.0.1]) by ms1.nttdata.co.jp (NTTDATA-TOP/DLV-20050621) with ESMTP id jA127ILb010383 for ; Tue, 1 Nov 2005 11:07:18 +0900 (JST) Received: from mail2.nttdata.co.jp ([163.135.10.22]) by ms1.nttdata.co.jp (NTTDATA-TOP/RCV-20050621) with ESMTP id jA127IkN010378 for ; Tue, 1 Nov 2005 11:07:18 +0900 (JST) Received: from mail2.nttdata.co.jp (localhost [127.0.0.1]) by mail2.nttdata.co.jp (NTTDATA-TOP/DLV-20050621) with ESMTP id jA127HEw027371 for ; Tue, 1 Nov 2005 11:07:17 +0900 (JST) Received: from msggw-r01.msg.nttdata.co.jp ([10.10.1.55]) by mail2.nttdata.co.jp (NTTDATA-TOP/RCV-20050621) with ESMTP id jA127Hdt027362 for ; Tue, 1 Nov 2005 11:07:17 +0900 (JST) Message-ID: <4366CE49.1000501@nttdata.co.jp> Date: Tue, 01 Nov 2005 11:09:13 +0900 From: Satoshi Nagayasu MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Subject: [linux-lvm] Creating snapshot causes processes hang? Reply-To: LVM general discussion and development List-Id: LVM general discussion and development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , List-Id: Content-Type: text/plain; charset="us-ascii" To: linux-lvm@redhat.com Hi all, I'm testing PostgreSQL(RDBMS) backup using LVM2 snapshot on RHEL4. After creating a snapshot, some processes (kjournald, one PostgreSQL backend and others) go to iowait status, and they never come back. (see pid 8105 and 2973 in below) In this situation, one PostgreSQL backend process is waiting in COMMIT processing(it generates fsync() on logical volume), and also kjournald is waiting something. There is no kernel oops, and the processors are not used. When I set an PostgreSQL option not to generate fsync on COMMIT, it looks work well. No process hangs. I guess some race conditions is occured around kjournald. Any comments and suggestions? Thanks. ------------------------------------ # uname -a Linux st17 2.6.9-5.ELsmp #1 SMP Wed Jan 5 19:30:39 EST 2005 i686 i686 i386 GNU/Linux # df Filesystem 1K-blocks Used Available Use% Mounted on /dev/sda5 20161172 6717648 12419384 36% / /dev/sda1 202219 11423 180356 6% /boot none 1037448 0 1037448 0% /dev/shm /dev/mapper/vg0-pgdata 1032088 515532 464128 53% /pgdata # ps ax PID TTY STAT TIME COMMAND 1 ? S 0:00 init [3] 2 ? S 0:00 [migration/0] 3 ? SN 0:00 [ksoftirqd/0] 4 ? S 0:00 [migration/1] 5 ? SN 0:00 [ksoftirqd/1] 6 ? S< 0:00 [events/0] 7 ? S< 0:00 [events/1] 8 ? S< 0:00 [khelper] 9 ? S< 0:00 [kacpid] 30 ? S< 0:00 [kblockd/0] 31 ? S< 0:00 [kblockd/1] 44 ? S< 0:00 [aio/0] 45 ? S< 0:00 [aio/1] 32 ? S 0:00 [khubd] 43 ? S 0:11 [kswapd0] 118 ? S 0:00 [kseriod] 186 ? S 0:00 [scsi_eh_0] 201 ? S 0:17 [kjournald] 1157 ? S