>
>
> I would slightly prefer if we discuss it over plain email (without
> patches), to see what you think about how complex the network model needs
> to be, and whether a static "time X" vs. semi-dynamic (based on the
> instance disk size) approach is best.
>
> Maybe there was some more
On 23 June 2016 at 17:42, Даниил Лещёв wrote:
> Hi, Iustin
>
>
>> Oh, no worries, I just wanted to know if Daniil acknowledged the comments
>> or not.
>>
>> Anyway, comments are welcome here and the discussion is still open:)
>>>
>>
>>
> The only reason why I didn't reply to
Hi, Iustin
> Oh, no worries, I just wanted to know if Daniil acknowledged the comments
> or not.
>
> Anyway, comments are welcome here and the discussion is still open:)
>>
>
>
The only reason why I didn't reply to your comments is that I wanted to
show patchset in order to discuss ideas in
On 23 June 2016 at 17:08, Oleg Ponomarev wrote:
> Hi Iustin, Daniil,
>
> The reason for Daniil not to reply immediately is his GSoC midterm
> evaluation coming soon. As the implementation represents his work during
> the first month of GSoC, it's necessary to share it with
> + improvement = (ini_cv - opt_cv) / (ini_cv - fin_cv - opt_cv)
The above expression is definitely incorrect (just consider opt_cv + eps =
fin_cv and fin_cv + eps = ini_cv). The improvement will be negative.
Do you mean "(ini_cv - opt_cv) / (fin_cv - opt_cv)" here?
> + in if long_sol_f ==
Hi Iustin, Daniil,
The reason for Daniil not to reply immediately is his GSoC midterm
evaluation coming soon. As the implementation represents his work during
the first month of GSoC, it's necessary to share it with the community at
this point. We have discussed your comments on design document
On 23 June 2016 at 16:45, wrote:
> From: Daniil Leshchev
>
> The patchset introduces new command line options
> (--long-solution-threshold" and --avoid-long-solutions").
> That gives an ability for HBal to avoid balancing solutions,
> that take significant
From: Daniil Leshchev
Change balancing functions in order to disallow long-time solutions,
whose K/N metric is less than algLongSolutionsFactor,
where K is solution's metric improvement and N is an estimated
time in seconds to perform this solution.
---
From: Daniil Leshchev
This tags make possible to specify network bandwidth between nodes.
The syntax is "htools:bandwidth:a::b::N", there N is network speed
between 'a' and 'b' tags.
---
src/Ganeti/HTools/Tags.hs | 60 +
From: Daniil Leshchev
This information is used by Hbal in order to prevent long balancing steps
("--avoid-long-solutions").
---
src/Ganeti/HTools/Loader.hs | 20 +++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/src/Ganeti/HTools/Loader.hs
From: Daniil Leshchev
These tests cover the following cases:
* default option values [1-3]
* avoiding slow networks [4-6]
* avoiding balancing moves at all [6-8]
All cases use inheritance of tags
---
test/data/htools/hbal-avoid-long-solutions.data | 17 +
From: Daniil Leshchev
The "long-solution-threshold" option specifies the threshold
in seconds, that defines long-time solutions.
The "avoid-long-solutions" option specifies the factor of avoiding
long-solutions. Allow only long-time solutions with K/N more than
avoiding
From: Daniil Leshchev
The patchset introduces new command line options
(--long-solution-threshold" and --avoid-long-solutions").
That gives an ability for HBal to avoid balancing solutions,
that take significant amount of time.
Daniil Leshchev (6):
Add
From: Daniil Leshchev
The bandwidths map store data about network speed
between current node and given Node by it's bandwidth tags.
---
src/Ganeti/HTools/Node.hs | 30 ++
1 file changed, 30 insertions(+)
diff --git a/src/Ganeti/HTools/Node.hs
On Thu, Jun 23, 2016 at 06:59:34AM -0700, David Mohr wrote:
>Hi Brian,
>this is the squashed commit of the actual code changes; without this
>the documentation update is slightly misleading ;-)
Woops. I guess I was a bit too quick off the mark!
>Thanks for applying it! I'll be
On Thu, Jun 23, 2016 at 03:57:50PM +0200, David Mohr wrote:
> The user id is used by ceph to determine the keyring to use for
> authentication. By default the admin keyring is used, which may
> not be desirable. Example usage:
>
> $ gnt-cluster modify -D rbd:user-id=foobar
>
> Signed-off-by:
Hi Brian,
this is the squashed commit of the actual code changes; without this the
documentation update is slightly misleading ;-)
Thanks for applying it! I'll be back soon with more patches.
~David
On Thursday, June 23, 2016 at 3:57:52 PM UTC+2, David Mohr wrote:
>
> The user id is used by
The user id is used by ceph to determine the keyring to use for
authentication. By default the admin keyring is used, which may
not be desirable. Example usage:
$ gnt-cluster modify -D rbd:user-id=foobar
Signed-off-by: David Mohr
---
lib/storage/bdev.py | 44
On Thu, Jun 23, 2016 at 12:57:01PM +0100, 'Federico Morg Pareschi' via
ganeti-devel wrote:
> When the watcher runs on each node group, if it can obtain the group
> lock, it submits a GROUP_VERIFY_DISKS job for each group. This happens
> every 5 minutes due to a cron job. This patch stops the
On Thu, Jun 23, 2016 at 02:05:30PM +0200, David Mohr wrote:
>
> Signed-off-by: David Mohr
Thanks. LGTM. I'll push it with some minor formatting tweaks.
Cheers,
Brian.
Signed-off-by: David Mohr
---
doc/design-ceph-ganeti-support.rst |9 -
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/doc/design-ceph-ganeti-support.rst
b/doc/design-ceph-ganeti-support.rst
index 7ec865c..9bd939c 100644
---
When the watcher runs on each node group, if it can obtain the group
lock, it submits a GROUP_VERIFY_DISKS job for each group. This happens
every 5 minutes due to a cron job. This patch stops the watcher from
submitting unnecessary verify disks jobs if there are some already
pending in the queue
22 matches
Mail list logo