[Gluster-users] Not real confident in 3.3

[Peter Linder]

For me, on wheezy, its here:

# cat /sys/kernel/mm/transparent_hugepage/enabled
always [madvise] never

Check that it's not set to [always].

'been stable for me around a year now.

On 6/19/2012 3:02 AM, Philip Poten wrote:
> Hi,
>
> 2012/6/18 Jeff Darcy<jdarcy at redhat.com>:
>> (1) There are known problems with the interaction between FUSE and
>> transparent hugepages
>> (https://bugzilla.redhat.com/show_bug.cgi?id=3D764964). =A0This could cau=
> se
>> one or more of your server processes to lock up.
> Does anybody here know how to disable transparent huge page migration
> on current debian squeeze kernels? I can not find the /sys path
> mentioned in the bug report:
>
> openvz01:~# find /proc/|grep hugepage
> /proc/sys/vm/nr_hugepages
> /proc/sys/vm/hugetlb_shm_group
> /proc/sys/vm/hugepages_treat_as_movable
> /proc/sys/vm/nr_overcommit_hugepages
> openvz01:~# find /sys|grep hugepage
> /sys/kernel/mm/hugepages
> /sys/kernel/mm/hugepages/hugepages-2048kB
> /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages
> /sys/kernel/mm/hugepages/hugepages-2048kB/nr_overcommit_hugepages
> /sys/kernel/mm/hugepages/hugepages-2048kB/free_hugepages
> /sys/kernel/mm/hugepages/hugepages-2048kB/resv_hugepages
> /sys/kernel/mm/hugepages/hugepages-2048kB/surplus_hugepages
>
> This is on squeeze kernel 2.6.32-5-amd64 with and without openvz patches.
>
> So, how can I implement the workaround from the bugreport? Is
> /proc/sys/vm/hugepages_treat_as_movable what I'm looking for? Because
> that's already 0.
>
> Any insight would be greatly appreciated, as glusterfs is a major pita
> with NFS due to the memory leak, and with FUSE due to the process
> hangs. I'd like to work around at least one of these problems.
>
> <rant>
> With normal use, glusterfs is factually unsuitable for production and
> only still in use here because we already have large amounts of data
> on it that are difficult to migrate.
>
> What irks me most is the "Priority: medium" and "Severity: low"
> classifications in that particular bug report. There doesn't seem to
> be any problem awareness :(
> </rant>
>
> That being said, I of course appreciate the work being done by the
> developers and hope there is a great future for the product.
>
> regards,
> Philip
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users