Hi Russell, On 10/02/2014 06:06 AM, Russell Smith wrote: > I behaved lazily... and have been trying to collect more data before > posting back. Sorry for the delay. > > 1. I can only reproduce the bug about once a week on 3.13 kernel that is > running on ubuntu 14.04. > 2. I was lazy and after reading patches and kernel stuff, I marked it as > fixed upstream. I didn't want to go compiling my own kernel as I need all > the virtio options.
Ubuntu provides upstream kernels here: http://kernel.ubuntu.com/~kernel-ppa/mainline/?C=N;O=D And some documentation about it is available here: https://wiki.ubuntu.com/Kernel/MainlineBuilds > I have 5 virtual machines running on top of a md10 + md1 raid all put > into a single lvm volume. The host is running ubuntu 12.04 with a 3.2 > kernel series. The guests are all virtio and running 14.04 as they have > been upgraded. Interestingly I also had an hypervisor running 12.04 and possibly with the 3.2 or the 3.11 kernel. IIRC, the guest was running a fresh install of 14.04. Since then, I reinstalled the hypervisor with 14.04 and the guest too. I never experienced this problem after this. > Most of the 5 instances have experienced a file system > read-only event. One server has experienced 3. It is my package mirror > running approx. IIRC, my guest was an apt-cacher-ng instance. > So after further investigation this may not be resolved upstream. If > you can provide me with a virtio/ubuntu compatible newer kernel I can > put it on a machine and try it out. But because I can't reproduce this > at will it's very hard to give a reproduction case. Indeed but since I haven't reproduce this issue once I'm afraid I can't help you here. > I found http://lists.openwall.net/linux-ext4/2014/05/21/2 which > indicates there is a bug somewhere if we are seeing failures where we > are. The result of that thread was more debugging put into the section > of code that reports this. I'm happy to track and patch any amount of > debug. > > As the filesystem re-mounts readonly, I suspect I should move /var/log > onto another filesystem to increase the chances I'll log everything. You can also send the logs to a remote syslog server. Good luck and thanks! Simon -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1298972 Title: stacktrace in ext4: /build/buildd/linux-3.13.0/fs/ext4/ext4_jbd2.c:259 __ext4_handle_dirty_metadata+0x1a2/0x1c0() To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1298972/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
