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Also, the check for the number of nodes is obsolete, as we compute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>

The “constants” module is a bit special in the sense that we don't want
to export random stuff from it. This unittest checks the naming
convention and removes imported modules from the module's namespace.

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

This patch removes all remains of the HTTP client pool. Newly added unittests
provide 96% coverage on http.client.

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

The HTTP pool is no longer used.

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

* devel-2.5:
  rpc: Disable HTTP client pool and reduce memory consumption
  Preserve bridge MTU in KVM ifup script
  hail: Fix result for node evacuation
  Fix assertion error on unclean master shutdown

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

Currently, we get a luxi Client, then parse the filter, then execute
the query. If parsing the filter fails, we connected to the masterd
needlessly.

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

* stable-2.5:
  rpc: Disable HTTP client pool and reduce memory consumption
  hail: Fix result for node evacuation
  Fix assertion error on unclean master shutdown

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

Currently, we have 4 types of arguments in LUXI calls:

- most common, a list of values
- a single argument that is sent as a list of one element
- a single argument that is sent by itself
- a dictionary (only Query and QueryFields)

This inconsistency makes it not only harder to auto-generate the
HTools LUXI interface, but also in general to check the arguments and
(if we ever want to do it) auto-generate the Python LUXI client.

Compare this with the node daemon, which uses consistently a list for
its arguments, and even with way more changes over time had no issues
with extending the interface.

In case we want to extend a call, there are two options:

- preferred: add a new call, keep the old one unchanged
- possible: add further parameters to the current argument list

The patch against HTools will follow—sending separately as the Python
changes are very clear by themselves.

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

We currently use 'filter' as the OpCode, QueryRequest and RAPI field
name for representing a query filter. However, since 'filter' is a
built-in function, we actually have to use filter_ throughout the code
in order to not override the built-in function.

This patch simply goes and does a global sed over the code. Due to the
fact that the RAPI interface already exposed this field, we add
compatibility code for now which handles both forms.

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

* devel-2.4:
  rpc: Disable HTTP client pool and reduce memory consumption
  Fix assertion error on unclean master shutdown

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

We noticed that “ganeti-masterd” can use large amounts of memory,
especially on large clusters. Measurements showed a single PycURL client
using about 500 kB of heap memory (the actual usage depends on versions,
build options and settings).

The RPC client uses a per-thread HTTP client pool with one client per
node. At this time there are 41 non-main threads (25 for the job queue
and 16 for client requests). This means the HTTP client pools use a lot
of memory (ca. 200 MB for 10 nodes, ca. 1 GB for 50 nodes).

This patch disables the per-thread HTTP client pool. No cleanup of
unused code is done. That will be done in the master branch only.

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

Untill now htools did not have support for generic Query in Luxi. This
patch introduces Query as a supported Luxi operation and replaces
QueryNodes, QueryInstances and QueryGroups with Query.

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

This patch simplifies the generation of save constructors for LuxiOp
by always using showJSON over an array of JSValues, instead of having
to pass showJSON in most cases, except the 5-tuple case.

Signed-off-by: Agata Murawska <agatamurawska@google.com>
Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>
[iustin@google.com: fixed a few issues]

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

This is not complete, but is as close as I can get it for now. I
expect people actually implementing the various changes to extend the
design doc.

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

Closes: #201 - KVM_IFUP does not set bridge-MTU on tap devices
Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This was, AFAIK, never used, and complicates the output code enough
that it's better to remove it.

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

This is just moving code around. A subsequent patch will do a bit more
cleanup and changing the output.

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

This is just beautification when dumping splices to stdout, as ghc
will optimise the 'id' away anyway.

Original generate code:

  opToArgs QueryTags kind name = J.showJSON (id kind, id name)

Afterwards:

  opToArgs QueryTags kind name = J.showJSON (kind, name)

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

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Update the man page of gnt-cluster to contain the documentation of the
--ignore-errors and --error-codes verify options. Also, include the list
of the error codes and their documentation.

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

lib/constants.py
* add to each CV_E* tuple the documentation of the error code
* add the DOCUMENTED_CONSTANTS constant for the doc preprocessor

autotools/docpp
* add a new directive class CONSTANTS_<kind>, that gets data from
  constants.DOCUMENTED_CONSTANTS

lib/cmdlib.py
* modify the code that unpacked the CV_E* tuples to ignore the
  documentation parameter

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

autotools/docpp
* handle generic custom directives in the form <class>_<kind>
* adapt handling of query fields

build/sphinx_ext.py
* add the BuildValuesDoc function to output definitions using the sphinx
  syntax that was already used for query fields
* adapt BuildQueryFields to use BuildValuesDoc

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

According to the iallocator documentation the “node-evacuate” call needs
to return a list of jobs, not a list of lists of jobs.

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

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

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

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

Treat the gnt-cluster verify errors identified by the error codes in
--ignore-errors as warnings; just print a warning message for the user.

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

lib/cli.py
- add IGNORE_ERROR_OPT;

client/gnt_cluster.py
- pass the ignore_errors parameter to the opcodes

lib/opcode.py
- update OpClusterVerifyConfig, OpClusterVerify and OpClusterVerifyGroup
  to accept the ignore_errors parameter

lib/cmdlib.py
- pass the ignore_errors parameter to the opcodes that need it

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

- move the cluster verify error codes from cmdlib._VerifyErrors to
  constants;
- add to each of them the CV (Cluster Verify) prefix;
- add the CV_ALL_ECODES and CV_ALL_ECODES_STRINGS constants;
- wrap the lines that exceed 80 characters after changing the error
  code names to the new ones.

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

Change 5a8648eb changed the order of the
return values of backend.GetMasterInfo(). This broke the users of the
master_info RPC.

This change restores the original order, and adds a comment in
bootstrap.py about the new value added to the return values of
master_info.

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Add the master_netmask cluster parameter, that represents the netmask of
the master IP, encoded as a CIDR suffix.

This parameter can be set via the --master-netmask of gnt-cluster init
and gnt-cluster modify. The default behaviour is to be consistent with
the old default (/32 for IPv4 and /128 for IPv6).

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Add the following methods to netutils.IPAddress:
* ValidateNetmask
* GetClassFromIpVersion
* GetClassFromIpFamily

Also, add related tests to the test suite.

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

* devel-2.5:
  cluster-merge: log an info message at node readd
  Bump version to 2.5.0~rc1
  Fix issue when verifying cluster files
  Revert "utils.log: Write error messages to stderr"
  Fix adding nodes after commit 64c7b383
  LUClusterVerifyGroup: Spread SSH checks over more nodes
  Optimise cli.JobExecutor with many pending jobs

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Andrea Spadaccini <spadaccio@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

node readd can take a long time, it's good to have info messages to see
progress.

Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Andrea Spadaccini <spadaccio@google.com>