Currently we don't set a boundary on the length of the score field,
which breaks all our nicely-aligned output efforts.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Currently Cluster.printSolutionLine function always wants to print the
old and new secondary, and as such it fails (ugly) for moves of
non-DRBD instances. Change it so that it looks explicitly for
noSecondary nodes, and format the output different in that case.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Currently we track the mirroring type as simply yes/no, with "yes"
meaning DRBD. This is not enough for handling shared storage, so we
need to add a type that gives the exact mirroring type (none,
internal, external), and a function that ties the disk template type
to the mirroring type.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Instead of computing the bad/good node lists separately (in effect
iterating twice over the all_nodes list and also creating that list
twice from the container), let's use 'partition' and do a small
improvement.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Currently, the code in Loader.hs marks as un-movable any instances
which don't have a secondary. This is not valid for any shared storage
templates, so let's remove that check completely. The initial state of
the instance will reflect correctly the movable status, based on the
disk template of the instance, so we don't need to re-do this check.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Our current cluster model is based on per-node storage, and as such we
don't handle shared disk templates. Until we can improve the model,
let's handle nodes without local storage better: in order to still
compute a valid cluster score, we need to handle nodes with tDsk == 0.

For this, we abstract the in-line computation of pDsk (which is used
in the cluster score) to a separate function which, for tDsk == 0,
returns instead 1 (i.e. disk fully used).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This make it possible to load from an Ialloc file

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This makes it possible to reuse that function when we want to use IAlloc
as a data source backend

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This helper abstract away all the unlines and unwords call you've to do
after you got a formatTable back.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This fixes the prop_Node_rMem test, which cares about memory only;
otherwise we could fail due to other problems (e.g. not enough disk).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This is not perfect, as the override applies identical target test
values to both the 'fast' and 'slow' tests (making the slow tests take
a long time), but it allows a quick override for manual runs.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

These just test that add/remove are idempotent. This is not perfect,
as we use unsorted lists for some values (instead of sets), so when
using non-empty nodes this would break (but for empty nodes, a
1-element list is sorted, so it's fine).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This makes balancing use the new metric.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This simply tracks the instance-used spindles (using hard-coded '1'
per instance), and additionally prevents additions in soft-mode when
we go over the limit.

Note: there's an assymetry between addSec and removeSec (basically
secondaries should only exist with DTDrbd8, so the check for uses_disk
is superflous), will be cleaned up later.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Since we use the functions always module-qualified, let's drop the
`instance` prefix and use a shorter one.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This patch adds two helper functions and uses them for a somewhat
simplified logic in the add/remove pri/sec node functions.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

If missing, we'll default to 1 spindle.

This also updates the text file format with the ipolicy, which was
forgotten.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This adds the spindleCount, hiSpindles and instSpindles attributes.

The spindleCount is equivalent to spindle_count on the ganeti side (a
node parameter). hiSpindles is the maximum instance-used spindles, and
instSpindles will be used to track spindles as used by instances (in
later patches).

The patch also reads spindleCount from the live backends. The text and
simu backends currently set it to 1, hard-coded.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Just the usual data/type declarations, read/save in the text backend,
etc. Also does a bit of unit-test cleanup (the way we build
ipolicies).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Also rename some functions as they don't reflect their actions anymore

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

It prints the information which hbal prints in a nice formatted manner and of
course without doing any actions. An example output looks like this:

$ hinfo -m xen.example.com
Loaded 4 nodes, 2 instances
Cluster has 1 node group(s)
Cluster coefficients:
   Field                Value Weight
   free_mem_cv     0.00195342  x1.00
   free_disk_cv    0.00000000  x1.00
   n1_cnt          0.00000000  x1.00
   reserved_mem_cv 0.00195342  x1.00
   offline_all_cnt 0.00000000  x4.00
   offline_pri_cnt 0.00000000 x16.00
   vcpu_ratio_cv   0.12500000  x1.00
   cpu_load_cv     0.50000000  x1.00
   mem_load_cv     0.50000000  x1.00
   disk_load_cv    0.00000000  x1.00
   net_load_cv     0.50000000  x1.00
   pri_tags_score  0.00000000  x2.00
Cluster score: 1.62890685

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Notes:
  - it'd be nice if there was a way to automatically generate the
    DiskTemplate list instead of manually specifying it. After all we
    have C.diskTemplates
  - Of course this actually does nothing, as for other
    "half-implemented" templates :(

Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Sorry, I missed this when I wrote the code originally.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

When using a really random "extra cpus" variable, this can overflow
(especially on 32 bit) and thus the final value can be < 0, thus the
test will fail to fail.

It doesn't make sense to test with huge values, so let's just generate
a reasonable value.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This was missing before, so with the new metric (NCpus), it's a good
time to test this.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This patch changes the allocation delta stats from RSpec to
AllocInfo. There's nothing interesting about it, but it should allow
more correct computations of Npus in hspace.

I also moved the AllocStats type alias from Cluster.hs to Types.hs
(just makes more sense).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

The current RSpec type doesn't cut it anymore (and it's used in other
places), so let's introduce a separate type for holding these
statistics.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This will be used in hspace for proper tracking of used/pool/unavail
npus.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Otherwise the docstrings for members take too many lines.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This is superseded now.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Instead of always overriding, we should update the nodes's maxcpu
setting only when the option is specified. Otherwise this will be read
from the cluster or from the Ganeti constant for non-live backends.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

hspace computes the NPU (normalised CPU units) via the
(used|pool|avail) VCPU divided by the (currently cluster-wide) max
vcpu_ratio.

However, in the future we'll have variable vcpu_ratios, so we can't
the calculation as such. We temporarily change to computing the
average vcpu_ratio across the cluster, however long-term this needs to
be improved, as ∑ (Vi/Pi) ≠ ∑ Vi / ∑ Pi. In the meantime we implement
this patch to remove the use of a global mcpu variable.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Since the iPolicy contains the vcpu_ratio now, we can remove the
per-node mCpu value (which was the initial way to set this attribute).

Furthermore, we need to re-set the hiCpu value when changing the
policy.

And of course some associated unittest changes.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>