On Sat, 21 Apr 2012, Wido den Hollander wrote:
> Hi,
>
> I was just playing with my libvirt driver and I noticed a stall.
>
> On one host I had running:
>
> for i in {0..8192}; do rbd create --size 64 image-$i; echo "created image-$i";
> done
>
> And on the other host I was running a pool-refre
On Sat, 21 Apr 2012, Andrey Korolyov wrote:
> Hello everyone,
>
> I have just tried ceph collectd fork on wheezy and noticed that all
> logs for ceph plugin produce nothing but zeroes(see below) for all
> types of nodes. Python cephtool works just fine. Collectd run as root
> and there is no obvio
On Sat, 21 Apr 2012, wrote:
> Oh, perhaps I find the reason for the bad behavior.
>
> According to Jenkins hash algorithm, any number of lower bits of the
> hash result is uniformly distributed, but the modular arithmetic can
> destroy the uniform distribution, unless the size of the bucket is th
On Sat, 21 Apr 2012, wrote:
> Hi, everyone.
>
> While reading crush_remove_tree_bucket_item, I found that removed
> items will still occupy its original position in the tree, but set its
> weight to 0.
Yes.. moving items around can result in shifted data.
> And in crush_add_tree_bucket_item, wh
Hi, everyone.
While reading crush_remove_tree_bucket_item, I found that removed
items will still occupy its original position in the tree, but set its
weight to 0.
And in crush_add_tree_bucket_item, why don't we reuse these items with
0 weight since I don't see additional data migration if we reus
Hi,
I was just playing with my libvirt driver and I noticed a stall.
On one host I had running:
for i in {0..8192}; do rbd create --size 64 image-$i; echo "created
image-$i"; done
And on the other host I was running a pool-refresh in libvirt.
As I was tailing the libvirt log I noticed that
Hello everyone,
I have just tried ceph collectd fork on wheezy and noticed that all
logs for ceph plugin produce nothing but zeroes(see below) for all
types of nodes. Python cephtool works just fine. Collectd run as root
and there is no obvious errors like socket permissions and no tips
from its l
Oh, perhaps I find the reason for the bad behavior.
According to Jenkins hash algorithm, any number of lower bits of the
hash result is uniformly distributed, but the modular arithmetic can
destroy the uniform distribution, unless the size of the bucket is the
power of 2!
Am I right?
On Tue, Apr