These updates and clarifications are still useful.

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

It is not used anywhere.

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

Empty lines shouldn't get indented. Unittest included.

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

We don't re-use cURL objects anymore, so there's no need to reset them.
PycURL 7.19.0 has a reference counting bug leading to a crash after
a certain number of performed requests.

Since a unittest depended on this function for a test, it is replaced
with a per-request completion callback, which could also be used by
users of the HTTP client.

Unittests are updated and a test for the request object's “nice name”,
used by the lock monitor, is included.

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

While we won't use this function in Ganeti (yet), users should be aware
of the potential crashes caused by this functioni (discussed offline).
If the installed version of PycURL is not affected, this test is a
no-op.

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

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

* devel-2.5:
  Allow per-hypervisor optional files
  Add hypervisors ancillary files list
  xen: abstract a few hardcoded strings as constants

Conflicts:
	lib/cmdlib.py
	  - trivial

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

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

Rather than just allowing files for all nodes to be optional, we allow
optional files to be per-category. The way this works is that they must
be included in both lists (the new code checks for this).

The code also removes a duplicate assert (present both in verify and
compute ancillary files) and checks the new functionality in unittests.

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

These lists will be used to declare some of the files not necessarily
needed on all nodes. The files selected are files without which the
various hypervisors can still work, but that when they are present
should be synchronized across the cluster (or node group).

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

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

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

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

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

The rename is not needed either, since we're not adding more code as of
now.
This reverts commit 57ca011e.

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

We discussed that this is not needed right now, and it breaks existing
functionality and unittests.

This reverts commit 6915fe26.

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

In python 2.7 the ovf unittests fail because OVFReader expects
ElementTree.parse() of an erroneous document to throw an
xml.parsers.expat.ExpatError while instead it throws an
ElementTree.ParseError.

The solution is to "except" for both errors, with the catch that
ParseError didn't exist before, so we need to define it locally and
get it from the module if it exists, while leaving it set to "None"
(thus catching no exception) if it does not.

Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Agata Murawska <agatamurawska@google.com>

* devel-2.5:
  Revert "rapi.client.ModifyNode should PUT rather than POST"
    - also fix the actual call, which was merged as PUT in master by
      mistake.

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

This was a mistake on my side because ModifyGroup and ModifyInstance
were PUT, and I was not aware of the discussion and the rationale why
this one had to be POST.

This reverts commit 55ef0cf6.

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

* devel-2.5:
  Revert "Added SPICE TLS option and related cert paths"
  Revert "Implementation of TLS-protected SPICE connections"
  Revert "Updated man pages with new SPICE TLS options"
  Revert "Add tls_ciphers and use_vdagent options"

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

This reverts commit bfe86c76.
This commit will be readded on master.

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

This reverts commit b6267745.
This commit will be readded on master.

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

This reverts commit b8a10435.
This commit will be readded on master.

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

This reverts commit 3e40b587.
This commit will be readded on master.

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

* devel-2.5:
  rapi.client.ModifyNode should PUT rather than POST
  Fix RAPI node modify client and server calls
  xen: changes to facilitate "xl" support (xen 4.1)
  xen: abstract instance config file naming
  Abstract xen's 'xm' command as a constant
  Fix RAPI documentation build
  rapi: Allow auto-promotion on node role change
  rapi: Add resource for modifying node
  opcodes: Add comment to *SetParams result description

Conflicts:
	lib/rapi/client.py
              - both functions stay, remove one empty line
	lib/rapi/rlib2.py
              - convert new function to 2.6 rapi style
	test/ganeti.rapi.client_unittest.py
              - both tests stay, trivial

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

This was caught (albeit in a sibylline manner) by unittests on master
which are not present in 2.5.

Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

rapi.client.ModifyNode accepts a "group" and not a "node" param.
(this bug is invisible but still not nice)

rlib2.R_2_nodes_name_modify submits the opcode with instance_name rather
than node_name as a param. This would break the call.

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

- Copy the xl config file, in case there's any
- Start instances by config file, not name (also xm compatible)
- Start paused domains with -p and not --paused (also xm compatible)
- Add a fixme for migration (changes are not xm compatible)

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

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

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

- Split code using ctypes directly into a helper class
- Don't load “libc.so.6”, but use handle for main program instead (see
  comment in code)
- Clarify comment on errno with older ctypes versions
- Rename unittest since it can't be used for other functions (modifies
  process environment at runtime)
- Add boolean return value for “Mlockall”

These changes are leftovers from some experiments with ctypes.

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

Renaming so that more code using ctypes could be added to the same file.

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

I am currently able to run a 2-node virtual cluster on my machine,
with a very ad-hoc setup. But the results show clearly that this is
doable, and that given the right tools, setting up such a cluster will
be quite easy.

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

This improves devnotes.rst with some tricks for Haskell development,
and additionally it does two Makefile improvements:

- properly document lib/_vcsversion.py as a requirement for
  Constants.hs (but do not require rebuild when updated)
- move HEXTRA at the end of the GHC invocation, so any command line
  options will indeed override the built-in ones (especially -osuf)

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

This moves the checking of results from the allocation functions to a
separate function, so that we have less code duplication. It also does
a bit of simplification in the printing functions.

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

The node offline/mcpu is identical to hbal's setNodesStatus, so let's
move that to CLI.hs and reuse it in hspace (also, rename it and drop
one 's').

Also, the check for the number of nodes is obsolete, as we compute
that from the disk template.

The patch does a bit of other small cleanups.

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

Both iterateAlloc and tieredAlloc share the same signature, but it's
not documented nor exported (needed for refactoring).

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

The logic is not entirely correct—the new Query interface exports the
field status, and we don't use that yet. But the new code should be
more readable.

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

This partially undoes commit 92678b3c, more specifically it removes the
Store data type and the associated code, since all Luxi arguments are
now lists.

Furthermore, since the qfilter field on Query is complex (it's
actually a tree structure), and we don't support it, turn it into a
plain () type, which always gets encoded as JSNull ('null'), so that
we can remove the optional field handling from Luxi (all fields are
always required).

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

*mumble*

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