[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Issues with devicemapper - advice for the others



The only production certified driver for docker at the kernel level is
device mapper, as Scott notes.  We at this time do not recommend
production workloads on btrfs or overlayfs, although overlayfs is
moving in a direction where it may be useful.  Note that overlayfs is
not posix compliant, so regular linux applications can and have broken
on it - you should not enable overlayfs with a long period of time
spent running your apps on it.

On Wed, Feb 10, 2016 at 11:47 AM, Scott Dodson <sdodson redhat com> wrote:
> Are you using direct-lvm or loop-lvm? Red Hat definitely considers
> devicemapper with direct-lvm to be production ready and we're keen to
> fix any issues you find with that setup. If your kernel panic was with
> direct-lvm a bug would be great; the backtrace, kernel version, and
> device-mapper-libs version should be enough to get us somewhere.
>
> If you need to configure direct-lvm and you're using RHEL/Centos
> packages you should have docker-storage-setup see this bit of
> documentation in the pre-requisites section
> https://docs.openshift.org/latest/install_config/install/prerequisites.html#configuring-docker-storage
>
> If you're not using centos/rhel packages give this a try, I've not
> followed this myself but it looks sane
> https://docs.docker.com/engine/userguide/storagedriver/device-mapper-driver/#configure-direct-lvm-mode-for-production
>
> On Wed, Feb 10, 2016 at 11:09 AM, Philippe Lafoucrière
> <philippe lafoucriere tech-angels com> wrote:
>> Hi,
>>
>> We have encounter a big issue with our cluster: We tried to run a graylog
>> server for a client inside a pod. The image graylog2/allinone seemed to be
>> (almost) perfect for that (see my post "Replica ID in env" in the dev
>> mailing-list).
>> After a few minutes, graylog stopped to processing messages. After a
>> discussion with their dev team, it appeared devicemapper was the source of
>> the problem.
>> We tried to increase the log level on graylog, and it generated even more
>> I/O, resulting in ... nodes kernel panic.
>> We are evaluating the possibility of using btrfs or overlay instead.
>>
>> Be careful with the ansible playbook, devicemapper is the default storage
>> driver on redhat oses, and it's considered NOT ready for production
>> (https://docs.docker.com/engine/userguide/storagedriver/selectadriver/). I
>> can confirm that today...
>>
>>
>> Cheers
>> Philippe
>>
>> _______________________________________________
>> users mailing list
>> users lists openshift redhat com
>> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
>>
>
> _______________________________________________
> users mailing list
> users lists openshift redhat com
> http://lists.openshift.redhat.com/openshiftmm/listinfo/users


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]