From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeremy Fitzhardinge Subject: Re: I/O related (?) domU crash on xen 4.0 + pv_ops Date: Tue, 08 Jun 2010 16:17:06 -0700 Message-ID: <4C0ECF72.3050509@goop.org> References: <4C0EA505.8050500@buzz-media.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4C0EA505.8050500@buzz-media.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Rick Boone Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org On 06/08/2010 01:16 PM, Rick Boone wrote: > Hey, > > I'm running into an issue with pv_ops kernel (2.6.31.13) and xen 4.0 - > my domU's are continually locking up, under heavy IO load. My issue > seems similar to what these guys are reportiing: > > https://bugzilla.redhat.com/show_bug.cgi?id=551552 > https://bugzilla.redhat.com/show_bug.cgi?id=526627 > https://bugzilla.redhat.com/show_bug.cgi?id=550724 > > Any solutions/ideas that haven't been covered in those reports? I've > turned off tickless kernel, and have also set the guest I/O scheduler > to "noop", but machines are still crashing. I'm using LVM-backed block > devices on the dom0. For awhile, I didn't have the kernel set to panic > on a hung task, and from digging around while the kernel was still up, > I was able to determine that the device that seems to be causing > issues is one that sees a lot of IO (it's receiving all of the logs on > a heavily used web server). Does booting the domU with "clocksource=jiffies" help at all? Have you tried a 2.6.32-based domU? I used to see these kinds of lockups with older kernels on my linode domU, but have seen none since moving it to Fedora 12 with a 2.6.32 kernel. Are you using a distro kernel, or one you've built yourself? J > > Here's a couple of my tracebacks: > 1) > > INFO: task pdflush:36 blocked for more than 120 seconds. > "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. > ffff8801e963f9c0 0000000000000282 00000000e4f10f56 ffff8801e963f9d0 > ffff8801eb7a31b0 ffff8801eb67c240 ffff8801eb7a3590 0000000103902b94 > 00000000e4f10f56 ffff8801e963fa70 ffff8801e963f9b0 ffffffff811f14ec > Call Trace: > [] ? blk_unplug+0x56/0x72 > [] io_schedule+0x37/0x59 > [] ? block_sync_page+0x5b/0x71 > [] sync_page+0x5a/0x72 > [] __wait_on_bit_lock+0x55/0xb3 > [] ? sync_page+0x0/0x72 > [] ? find_get_pages_tag+0xf7/0x144 > [] __lock_page+0x71/0x8c > [] ? wake_bit_function+0x0/0x51 > [] write_cache_pages+0x201/0x3bf > [] ? __writepage+0x0/0x5a > [] ? xen_force_evtchn_callback+0x20/0x36 > [] ? xen_restore_fl_direct_end+0x0/0x1 > [] generic_writepages+0x35/0x4f > [] do_writepages+0x3f/0x5e > [] writeback_single_inode+0x161/0x2d7 > [] generic_sync_sb_inodes+0x1ef/0x355 > [] ? pdflush+0x0/0x286 > [] sync_sb_inodes+0x1c/0x32 > [] writeback_inodes+0x79/0xdf > [] ? sync_supers+0xb3/0xce > [] wb_kupdate+0xb9/0x13a > [] ? pdflush+0x126/0x286 > [] pdflush+0x163/0x286 > [] ? wb_kupdate+0x0/0x13a > [] ? pdflush+0x0/0x286 > [] kthread+0x9e/0xa8 > [] child_rip+0xa/0x20 > [] ? int_ret_from_sys_call+0x7/0x1b > [] ? retint_restore_args+0x5/0x6 > [] ? child_rip+0x0/0x20 > 1 lock held by pdflush/36: > #0: (&type->s_umount_key#23){......}, at: [] > writeback_inodes+0x63/0xdf > Kernel panic - not syncing: hung_task: blocked tasks > Pid: 34, comm: khungtaskd Not tainted 2.6.31.13-xen-4.0.0 #4 > Call Trace: > [] panic+0xb2/0x168 > [] ? print_lock+0x96/0xb1 > [] ? lockdep_print_held_locks+0xa5/0xc9 > [] ? show_stack+0x2a/0x40 > [] ? touch_nmi_watchdog+0x6c/0x87 > [] ? __debug_show_held_locks+0x33/0x49 > [] watchdog+0x209/0x258 > [] ? watchdog+0x75/0x258 > [] ? complete+0x52/0x71 > [] ? watchdog+0x0/0x258 > [] kthread+0x9e/0xa8 > [] child_rip+0xa/0x20 > [] ? int_ret_from_sys_call+0x7/0x1b > [] ? retint_restore_args+0x5/0x6 > [] ? child_rip+0x0/0x20 > > --------------------------------------------- > > 2) > INFO: task kjournald:951 blocked for more than 120 seconds. > "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. > ffff8801c8993bf0 0000000000000282 00000000cf63a654 ffff8801c8993c00 > ffff8801ca899090 ffff8801ccae31b0 ffff8801ca899470 0000000000000001 > 0000000000000001 0000000000000200 0000000000000001 000000000160015f > Call Trace: > [] io_schedule+0x37/0x59 > [] ? blk_backing_dev_unplug+0x20/0x36 > [] sync_buffer+0x51/0x69 > [] __wait_on_bit+0x54/0x9c > [] ? sync_buffer+0x0/0x69 > [] ? sync_buffer+0x0/0x69 > [] out_of_line_wait_on_bit+0x7d/0x9e > [] ? wake_bit_function+0x0/0x51 > [] __wait_on_buffer+0x32/0x48 > [] journal_commit_transaction+0x684/0x12f2 [jbd] > [] ? check_events+0x12/0x20 > [] ? xen_restore_fl_direct_end+0x0/0x1 > [] ? _spin_unlock_irqrestore+0x44/0x5f > [] ? try_to_del_timer_sync+0x65/0x84 > [] ? del_timer_sync+0x0/0xa0 > [] kjournald+0x161/0x3ae [jbd] > [] ? autoremove_wake_function+0x0/0x5e > [] ? kjournald+0x0/0x3ae [jbd] > [] kthread+0x9e/0xa8 > [] child_rip+0xa/0x20 > [] ? int_ret_from_sys_call+0x7/0x1b > [] ? retint_restore_args+0x5/0x6 > [] ? child_rip+0x0/0x20 > no locks held by kjournald/951. > Kernel panic - not syncing: hung_task: blocked tasks > Pid: 34, comm: khungtaskd Not tainted 2.6.31.13-xen-4.0.0 #18 > Call Trace: > [] panic+0xb2/0x168 > [] ? lockdep_print_held_locks+0x46/0xc9 > [] ? show_stack+0x2a/0x40 > [] ? touch_nmi_watchdog+0x6c/0x87 > [] ? __debug_show_held_locks+0x33/0x49 > [] watchdog+0x209/0x258 > [] ? watchdog+0x75/0x258 > [] ? complete+0x52/0x71 > [] ? watchdog+0x0/0x258 > [] kthread+0x9e/0xa8 > [] child_rip+0xa/0x20 > [] ? int_ret_from_sys_call+0x7/0x1b > [] ? retint_restore_args+0x5/0x6 > [] ? child_rip+0x0/0x20 > > -- Rick Boone > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel >