I made a mistake when adding support for htools to run-in-tempdir, by
needlessly copying the htools binaries, when it's enough to link
them. The copying is only really needed for python code, so we change
the Haskell binaries to linking and also a few other directories.

Difference:
- before: 0.8s, 95MB copied
- after: 0.6s, 6.2MB copied

And, by the way, can't wait for Python 2.6 as minimum supported
version…

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

Add a new client called 'gnt-storage'.
The client interacts with the ExtStorage interface, similarly to
the way gnt-os interacts with the OS interface.

For now, only two commands are supported: 'info' and 'diagnose'.

'diagnose' calculates the node status of each provider on each node,
similarly to gnt-os diagnose. Furthermore, for every provider, it
calculates it's nodegroup validity for each nodegroup. This is done
inside the LU and not the client (marked as 'TODO' for the  global
validity of gnt-os diagnose).

In the future, gnt-storage can be used to manage storage pools,
or even be extended to diagnose other storage types supported by
Ganeti, such as lvm, drbd (INT_MIRROR) or rbd (EXT_MIRROR).

Signed-off-by: Constantinos Venetsanopoulos <cven@grnet.gr>
Signed-off-by: Iustin Pop <iustin@google.com>
[iustin@google.com: fixed Haskell compatibility and style fixes]
Reviewed-by: Iustin Pop <iustin@google.com>

This is a new personality that for the moment doesn't do anything.

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

Accidentally stumbled upon this while testing unrelated code on a
machine with ~3K active jobs - the bash completion unittest was
hanging.

Upon investigation, it turns out that bash's ${var//pattern/repl/} is
probably quadratic in the size of input (or worse, even):

  $ touch job-{1..500}
  $ time ( a=$(echo job-*); echo ${a//job-/}| wc -c; )
  1892

  real    0m0.597s
  user    0m0.590s

  $ touch job-{1..1000}
  $ time ( a=$(echo job-*); echo ${a//job-/}| wc -c; )
  3893

  real    0m4.654s
  user    0m4.580s

We can easily fix this if we change to array-based substitution (once
per element):

  $ time ( a=($(echo job-*)); echo ${a[*]/job-/} |wc -c; )
  3893

  real    0m0.028s
  user    0m0.010s

  $ touch job-{1..10000}
  $ time ( a=($(echo job-*)); echo ${a[*]/job-/} |wc -c; )
  48894

  real    0m0.233s
  user    0m0.220s

This means that exactly when the master node is busy processing many
jobs, we could accidentally start consuming lots of CPU in the bash
completion, which is not good.

Note: the code might have problems with filenames containing spaces (I
didn't reset the IFS, etc.), but the original code had the same issue,
I think.

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

This just ties in all the pieces so far and enables bash completion
for it.

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

This patch adds support for parsing the command list out of a binary
(very strict format), and then iterating over that and building the
sub-commands options/arguments.

It also does a bit of general cleanup in the script.

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

This single constant could be as well moved to constants.py. For now I
choose to just export the module as well.

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

There is no obvious need for this space--the line would wrap at the next
space, which comes soon after.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This commit adds shelltests for the mon-collector binary and for the DRBD
data collector.

Also, it fixes a small bug in the DRBD parser found thanks to the tests.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This patch enhances the opcode list checks - instead of spawning a
Python interpreter to display the opcode list, we export it statically
in Constants.hs via a slight convert-constants change.

Furthermore, since we now have opcode parity, we enable full opcode
list checks.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Signed-off-by: Apollon Oikonomopoulos <apollon@noc.grnet.gr>
Signed-off-by: Dimitris Aragiorgis <dimara@grnet.gr>
Reviewed-by: Iustin Pop <iustin@google.com>

Currently, the cluster verify errors are defined as follows:

CV_ECLUSTER_FOO = (TCLUSTER, "ECLUSTER_FOO", "description")

This means there's no standalone name for the string "ECLUSTER_FOO",
which makes it hard to derive automatically a type for this union in
Haskell.

There are three possible fixes:

- manually separate the CV_ECLUSTER_FOO_STR = "ECLUSTER_FOO" in
  constants.py
- manually extract the strings in Haskell code
- change convert-constants to automatically export virtual constants
  for those

After discussion on IRC, I've taken the latter approach; even though a
bit hack-ish, it avoids manual work and potential errors.

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

This is similar to commit f5ce7613. A stable sort order makes changes
between versions easier to find.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

- Debian Squeeze and up have a package named “python-coverage”, but it
  doesn't use the same binary name as upstream (“coverage”).
- Said package includes a patch to use symlinks instead of file copies
  for jQuery. If files from previous runs are around, an exception is
  raised. This is fixed by removing all regular files and symlinks
  before generating a report.

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

These will be needed for the serialisation of errors.

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

This part of the code was duplicated by accident.

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

This patch cleans up  the Haskell library detection and defaults.

First, it makes the base compiler/libraries required, per the email discussion.

It then adds two new small autoconf macros, on to check for a required
Haskell library and one to do custom action based on test results. We
use these macros to cleanup and simplify a bit the module detection:

- rapi, confd, and split query are auto detected and enabled if _all_
  required libraries are present
- unittests are enabled if _all_ required libraries are present

The patch also updates the documentation regarding required libraries.

After this patch, base Ganeti fully buildable on Debian Squeeze/Ubuntu
Lucid.

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

As I wrote during/after the review on commit 2958c56e, “ganeti-cleaner:
Separate queue cleaning code”, while I appreciated the permission
separation, I didn't like too much the file-based approach:

- it is a very simple script, and lots of the code is duplicated
  between the two; I wouldn't like to see "ganeti-vmcapable-cleaner",
  "ganeti-master-candidate-cleaner", etc. in the future
- ganeti-master-cleaner "pollutes" the namespace, creating
  tab-completion conflicts with ganeti-masterd

This patch simply merges the master-cleaner back into cleaner, while
keeping the separate user permissions scheme, separate log files, etc.

Additionally, it fixes two bugs in the unit-test (not run with set -u
and wrong path in the master-cleaner log files test; yay for even
worse safety than Python?).

And finally, since we have now support for --help-completion, it adds
bash completion support for this script :) (needs to be applied on top
of my argument support patch series).

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

This allows the command completion to treat them as separate option
types; this is not useful in build-bash-completion right now, but
might become so in the future.

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

This requires a few tiny changes to build-bash-completion and to
Makefile rules, but is straightforward.

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

Beside parsing the arguments as well, we also change the serialisation
format for choices/suggests, to make the Python code simpler.

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

Tests that depend on fakeroot were not distributed in the tarball created
by "make dist" when fakeroot was missing. Now all tests are distributed,
and a warning is printed when a test requires fakeroot but fakeroot is
missing.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

In older versions of GNU Bash extended patterns, such as “@(…)”, are only
available with the “extglob” shell option. As pointed out in [1] and [2],
“extglob” must be enabled while parsing the code. Therefore the flag must be
enabled at the beginning of the script and be reset to its original value at
the end as to not interfere with other code on shell initialization.

[1] http://unix.stackexchange.com/questions/45957
[2] http://mywiki.wooledge.org/glob



Reported by Sascha Lucas.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>
(cherry picked from commit 893ad76d)

This patch ties together the previous commits, by extending
build-bash-completion to call all htools scripts and get their
completion information, then use that to build fake cli_options
representing them and finally generate the bash completion
information.

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

In older versions of GNU Bash extended patterns, such as “@(…)”, are only
available with the “extglob” shell option. As pointed out in [1] and [2],
“extglob” must be enabled while parsing the code. Therefore the flag must be
enabled at the beginning of the script and be reset to its original value at
the end as to not interfere with other code on shell initialization.

[1] http://unix.stackexchange.com/questions/45957
[2] http://mywiki.wooledge.org/glob



Reported by Sascha Lucas.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Bernardo Dal Seno <bdalseno@google.com>

This reverts commit 94014b63. Turns out
I misunderstood the “extglob” shell option.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Bernardo Dal Seno <bdalseno@google.com>

First the numbers:
$ stat --format '%s %n' doc/examples/bash_completion*
77847 doc/examples/bash_completion
86492 doc/examples/bash_completion-debug

The non-debug version doesn't use indentation and does not have some
(primitive) facilities for debugging. The savings are about 8.5 kB or
10 %. The “-debug” version is used by “devel/upload”.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

In older versions of GNU Bash extended patterns, such as “@(…)” are only
available with the “extglob” shell option. This patch adds a wrapper
function so that extglob is always enabled while doing any completion.
Due to early returns inside the main completion function this seemed
like the best option.

Reported by Sascha Lucas.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Bernardo Dal Seno <bdalseno@google.com>

Future changes will change Path.hs to use an environment variable.

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

File system paths moved from constants to pathutils.

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

This moves the last (I think) htools-related bits out of test/ under
htest/.

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

This is the first commit of a series that will attempt to cleanup the
test code organisation, which evolved somewhat organically from the
initial pure htools functionality.

The proposed organisation of the tree will be as follows:

- htools (or maybe renamed to haskell or hs): only production code
- htest: top-level test directory, containing test.hs, static helper
  scripts, etc.
- htest/Ganeti/*.hs: modules implementing the actual test properties
  and test cases for the correspondingly-named production code modules
- htest/data: containing test data files for the test cases

This particular patch moves all the test code (test.hs, hpc-htools.hs
symlink) and helper scripts (offline-test.sh, etc.) from htools/ to
htest/, while updating the files themselves (if they had paths
mentioning htools/), .gitignore and the Makefile.

The only special mention is that in Makefile, we used to have a BINARY
shell variable in binary build rule; that was computed via stripping
`htools/' prefix; I've cleaned that and replaced with $(notdir $@);
even though it's duplicated a few times, it leads to more readable
make output (and easier to copy-paste).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

This enhances convert-constants to not flatten dicts completely, but
also generate a so-called association list for them.

This allows either direct use of the 'lookup' function, or (for
performance) conversion to Data.Map and optimised lookup later.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

While looking at the hypervisor types in Constants.hs, I saw that the
'hyperTypes' list is using strings instead of names. This is due to
the fact that we require the entire elements in the list to be
identified (homogeneous lists), but the string "xen-pvm" is declared
by both 'defaultEnabledHypervisor' and 'htXenPvm'.

We can improve convert-constants by adding a list of names that we
know (statically) don't declare, but instead reuse values. Unless we
have an ADT parser, this is the best we can do, I think.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

Since these will be needed for Haskell's implementation of query2.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

Debian Wheezy will ship with this version, and it has many improved checks compared to 0.6, so let's:

- bump version in the docs
- silence some new checks that are wrong due to our indent=2 instead of 4
- fix lots of errors in the code where the indentation was wrong by 1
  or 2 spaces
- fix a few cases of == True, False, None and replace with 'is'
- re-indent some cases where the code is OK, but pep8 complains

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

Currently, the order of commands in the bash completion file is
random, because the sub-commands are not sorted. This makes it harder
to investigate the differences in packaged Ganeti or in installed
Ganeti, since chunks in this file will have a random order.

To fix this, we sort the subcommands based on the first subcommand in
a given group ('first' also in sorted order); this results in a stable
contents of file, as tested by building it many times and checking for
differences.

The patch also does a few other minor changes to the file
(e.g. updating copyright years, etc.).

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

Allow all errors to be displayed, instead of aborting at the first
one, and don't show stacktraces anymore.

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

This is trivial, but should prevent such things in the future.

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

This adds a bit of dumb mapping of values to names, while trying to be
safe. This is the best we can do without resorting to parsing or
interpreting ASTs.

The difference in the output is:

 -- | Converted from Python list or set ADMINST_ALL
 adminstAll :: [String]
-adminstAll = ["down", "offline", "up"]
+adminstAll = [adminstDown, adminstOffline, adminstUp]

Since for most such values we use strings, we don't gain in type
safety on the Haskell side. But it makes the output more readable and
it might open up other opportunities later.

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