When slow clients post a large request, 30 seconds is too short for the current
  production usage. Short timeouts are rare but it always happens on POST, which
  is annoying for the user.

  Move from 30s to 5min the timeout to allow the slow clients upload their data
  with time.

  Apply commits:
   stack.erp5: Drop unneeded executable permissions. (fe7ea950)
   stack.logrotate: Fix support for stopped processes.
   stack.erp5: Use an iterator to produce port numbers.
   software/erp5: if wendelin-core-zblk-fmt is not given, then use wendelin.core's default.
   software/erp5 & stack/erp5: Add a new parameter wendelin-core-zblk-fmt.

In order to work in deep directory structures it is required to have
kvm-controller script wrapped to avoid shabang limit.

Since release of ipython 6.0.0 (
https://pypi.python.org/pypi/ipython/6.0.0 ) ipython became incompatible
with python < 3.3, so its version needs to be pinned.

Known nodejs bug[1] has proposed fix [2] applied directly here.

[1] https://github.com/nodejs/node/issues/9137
[2] https://github.com/riedonetworks/meta-nodejs/commits/dora/recipes/nodejs/files/0001-fix-execvp-printf-argument-list-too-long.patch

CLOUDOOO:
  Software Release: software/cloudooo/software.cfg
  Test Suite: CLOUDOOO-MASTER ran at 2017/04/18 09:23:8.409793 UTC
              0 failures, 2 errors, 200 total, status: FAIL

Failures on unit tests for software/cloudooo/software.cfg on the current revision:
  *  testPdfHandler 0 fails, 1 errors
  *  testPdfServer 0 fails, 1 errors

SlapOS Master
  Software Release: software/slapos-master/software.cfg
  Test Suite: SLAPOS-MASTER-MASTER ran at 2017/04/18 09:24:47.248034 UTC
              11 failures, 2 errors, 1316 total, status: FAIL

Failures on unit tests for software/slapos-master/software.cfg on the current revision:
  *  slapos_cloud:testSlapOSCloudAlarm 1 fails, 2 errors
  *  slapos_cloud:testSlapOSCloudConstraint 3 fails, 0 errors
  *  slapos_cloud:testSlapOSCloudDocument 1 fails, 0 errors
  *  erp5_web_shacache:testShaCache 1 fails, 0 errors
  *  slapos_erp5:testSlapOSERP5DefaultScenario 2 fails, 0 errors
  *  slapos_erp5:testSlapOSXHTML 2 fails, 0 errors
  *  slapos_monitoring_ui_test:testFunctionalSlapOSMonitoring 1 fails, 0 errors

ERP5:
  Software Release: software/erp5.software.cfg
  Test Suite: ERP5-MASTER ran at 2017/04/18 09:23:48.196234 UTC
              2 failures, 0 errors, 5402 total, status: FAIL

Failures on unit tests for software/erp5.software.cfg on the current revision:
  *  testFunctionalCore 2 fails, 0 errors

Wendelin:
  Software Release: software/wendelin/software.cfg
  Test Suite: WENDELIN-MASTER-DEV ran at 2017/04/18 13:08:15.910381 UTC
              0 failures, 0 errors, 10 total, status: PASS

NEO:
  Software Release: software/neoppod/software.cfg
  Test Suite: NEO-MASTER ran at 2017/04/18 14:01:23.263564 UTC
              16 failures, 1 errors, 803 total, status: FAIL

Failures on unit tests for software/neoppod/software.cfg on the current revision:
  *  MySQL 8 fails, 1 errors

/reviewed-on nexedi/slapos!159

It already happened that the first tests of the suite failed
because the MySQL server was not fully started:

    OperationalError: (2002, 'Can\'t connect to local MySQL server through socket \'.../var/run/mariadb.sock\' (2 "No such file or directory")')

  We require 8 spaces after the second line for adjust the layout at the generated file.

perl/bin/enc2xs can be a shell script that calls enc2xs.shebang so we
cannot call it explicitly with perl like "/path/to/perl
/path/to/enc2xs" but simply calling "/path/to/enc2xs" is enough and
safe.

  Make the integration with erp5testnode separeted from original SR.
  Optimise KVM resilient tests by downloading the test image on the software release site,
  without change the original SR. (This will sabe internet and speed up result).

also drop non-existing '--with-nm' option.

  The image still way too big but it is better the whole image

  This is just cosmetics.

  This fixes bugs on the order of parts and duplicated sections introduced on:
    8d28b7e0
    d3df3b6e

  Simple Don't repeat yourself.

  This change allow us simplify accross the instances usage of logrotate

  If the password file is overwriten by a third part application,
  an wrong n can be added to the file. The change on this recipe
  allow us automatically fix this miss behaviour.