Currently, the tryAlloc and tryReloc function return a list with all the
results, both failures and successes. This is fine for hail, which does
one round of allocations, but is not so good for hspace, which does
iterative rounds; since at each (successful) step we only take the best
solution, it means that we're using lots of heap space to compute and
store node lists which are thrown away at the end of the step.

This patch changes these two functions and their callers in hail/hspace
to only return the best solution, and error/success counters. This
allows hspace to run in a much smaller space, and reduces GC cost
greatly.

Overall, it is a cleanup, as hail/hspace did a lot of work to chose this
best solution, whereas now it's automatically promoted within
Cluster.concatAllocs.

Currently the Cluster.AllocSolution type is defined as a list of
‘(OpResult Node.list, …)’ and the results for applyMove are defined as
‘(OpResult Node.List, …)’. Both these means that the failure/success
indication is hidden in the first elements of this tuple, which makes is
harder to add other elements depending on the success/failure (like the
score for the new node list).

This patch moves the OpResult to outside of the tuple, in effect making
all these ‘OpResult (…)’ which makes the internal tuple elements
consistent. The patch is affecting the applyMove, tryAlloc and
tryRealloc functions, and only briefly the hail and hspace programs.

Currently many of the exit and warning conditions mistakenly display error
messages on stdout, which makes parsing the output of programs harder. This
patch attempts to fix such occurrences.

This big patch cleans up the code per hlint indications. Many removals
of extra parentheses, replacements of concat . map with concabtMap,
extra dollar signs, eta reductions, etc. were performed.

The code still compiles and passes a couple of manual tests on sample
files. The individual changes are also small enough as to be visually
easy to confirm.

Currently the allocation/move operations workflow return ‘Maybe a’,
which is very convenient but loses all details about the failure mode.

This patch introduces a new data type which encodes the specific failure
mode. It is not yet used correctly (e.g. all node operations result in
FailN1), but the workflow is updated. Most of the changes are in the
hail/hspace tools, the library code required only trivial adjustments.

This removes some unused functions and imports to cleanup the warnings.

In a few places (e.g. tryRead or any printf call) it's a little bit hard
to add the correct type signatures, but in the it is possible to fix
these warnings (which can bite one in subtle cases).

Currently tryAlloc/tryReloc do not return the new instance, as this is
not needed for IAllocator alloc/reloc requests. However, for computing
the space, the new instance is useful, so we modify these functions to
return this information too.

The patch also improves hspace to have (with default parameters) a
parseable output.

This doc-patch adds copyright and license information to (hopefully) all
needed files.

These are generic enough to be used from multiple places, they belong
better in Cluster.hs than in the hail source.

These two will be more generic than now, and belong somewhere else -
Loader.hs is a generic module for data loading, thus we move them there.

This is a first step towards full datatype renaming. That requires more
changes, so at first we only want to document clearly what is a node
index, what is an instance index, and what is a plain Int.

This patch makes the Types module a base module, and Node/Instance ones
import it, from the previous (opposite) situation. This will allow in
the future to use newtypes for the index and name types.

This patch fixes the relocate-to-primary-node problem, and doesn't mark
the excluded nodes as offline anymore (we don't use them anyway, so
let's not use a different/fake cluster state).

This patch implements filtering out of the offline/drained nodes and
fixes a bug in IAllocator.hs parsing (similar to an older bug in Rapi.hs
from where the code was copied).

This patch implements non-mirrored instance allocation, by allocating as
secondary node “noSecondary”.

Both allocate and relocate compute new node lists, whose score must be
ranked and the best option chosen. This means we can convert the code to
a generic function.

This patch implements allocate for two node requests. One node requests
can be done as soon as we have a valid allocateOn function for single
nodes.

This patch completes the implementation of hail relocate. It maps all
valid destination nodes through a ReplaceSecondary IMove, filters out
the failed relocations, computes the resulting scores and picks the
lowest one.

This patch implements a very stupid (and broken) version of hail
‘allocate’.

This is not yet used.

Since we have Node/Instance.name, we can now simplify a few constructs.

This patch generalizes a little the CLI handling by not passing in a
special function for help and such, but instead requiring that the
options object supports some common functionality.

This new big patch does a couple of more cleanups in the loading of data
chapter:
  - introduce a Types module that holds most types (except the base
    Node/Instance/etc.) so that multiple other modules can use these
    (instead of only Cluster and its users)
  - bring IAlloc.hs in line with the recent changes of providing data
    types and not strings
  - removal of obsolete Utils.hs json-related functions

This makes hail compile and get a request parsed via IAlloc, but nothing
more.

This patch changes Cluster.loadData to return a Result, instead of
directly the values; this will allow us to return meaningful error
values (e.g. when an instances lives on unknown node) rather than simply
abort. Currently the result is always an Ok, the actual signalling of
errors will come later.

Its callers are changed to accommodate for the new return type and to
display errors as needed.

This big patch converts from our home-grown monad-like constructs
(the Either stuff) to a real, Either-like-but-another-name monad.

We introduce a “Result a” monad, and this allows dropping many of the
extra constructs. Hopefully the code is also more clear.

Many of the functions could now be written in a generic-monad style,
instead of Result specifically, but that will come in future patches.

IAlloc.hs also has some unrelated patches.

This option is the opposite of the --verbose option, and it allows
decreasing the verbosity level from the default of one to zero (which
currently doesn't show the warning messages for missing disk/memory).

This patch moves the oneline format into a separate function for easier
usage.

For clusters with no instances, there is no point in computing either
the score or in running the algorithm. In this case, we exit prematurely
and when running in one-line mode we show dummy information.

This new parameter causes the algorithm to finish (or even not start at
all) if we reach/have a score better than it.

This patch adds support for selecting the instance/node file names via
two environment variables (HTOOLS_NODES, HTOOLS_INSTANCES).

Unfortunately we still have lots of duplicated code, since the options
are not unified.

Neither hbal nor hn1 take any arguments beside the options, so if any
are passed is most likely an unintended error. This patch aborts in such
cases.

This patch adds support in hbal for writing the command list to a shell
script, with error checking and allowing for early exit.

Since it's easy to pass a wrong node name as offline, we should abort
instead of silently ignoring it.

This patch adds checking of cluster data in the binaries and display of
node's x_mem/i_mem in the node list.

This patch moves the common CLI functionality (as much as currently
possible) into a separate module. This means we only have one parseOpts
and that Utils.hs doesn't keep this kind of functions anymore.

This is added for better readability, since this is very often used in
declarations.

This patch modifies hbal (only, hn1 not yet) for setting nodes offline.