Publish first class objects involved in an operation (for now vm uuid) on the event bus . Example - during attach/detachIso along with iso id, vm id should be available as well.
- Refactor tests:
- Upgrade tests to use jUnit4
- Add hamcrest dependency (contribution of by Laszio Hornyak)
- Break big tests in small unit tests
- Replace assertTrue/False with complex conditions by assertThat with
specific matchers
- Remove dead code:
- Private static method never called locally
- Add test for method that validates cidrs
Signed-off-by: Laszlo Hornyak <laszlo.hornyak@gmail.com>
encoded. This cause createStoragePool or addImageStore command to fail if special
characters were present. Updated the code to pass user, password and domain as part
of details while adding primary or secondary. Also made changes on server side to
handle it.
- StringBuffer replaced with StringBuilder
- nullcheck of tokens array removed since String.split does not return null
- unit test added
Signed-off-by: Laszlo Hornyak <laszlo.hornyak@gmail.com>
For some reason the mgmt server is required to have access to the template
being registered in order to check its size against resource limits
during registration. This is bad, but worse, it's a sync call, so this
call can hang for up to the default HTTP connection timeout.
Reducing http timeout as a quick workaround, but this check needs to
1) be removed or 2) done later or 3)registerTemplate needs to be async
and 4) moved off the mgmt server (ssvm is the only thing that needs access to
template url).
This patch adds support for trust chains in the netscaler.
I initially planned on using the 10.1 API's "bundle" feature but during
my testing I found that was not working. So I am doing the chain linking
myself. Also NS can have only one entity of a certificate ie lets say
two different users try to add the same certificate on the netscaler
only one of them will go through. The other one says resouce already
exists even though they have different files.
This can be a problem in trust chains where the chain can be shared
between multiple accounts/certificates. So, I am using the figerprint as
an identifier of a certificate and making sure that we delete it only
when no one references it.
- new utility method introduced in PropertiesUtil to load properties objects from files
- RegionManagerImpl modified to use the utility method
- Tests added for both
Signed-off-by: Laszlo Hornyak <laszlo.hornyak@gmail.com>
Supplementing the fix/commit 748315620b
While remote executing commands through ssh, wait for channel condition of EXIT_STATUS.
Signed-off-by: Sateesh Chodapuneedi <sateesh@apache.org>
Currently during ssh execution of remote command, if no response is received within timeout, Cloudstack is returning success result.
This is resulting in false positives. Fix is to check if exit status of remote command is available or not. If not, return failure result.
Signed-off-by: Sateesh Chodapuneedi <sateesh@apache.org>
Changes:
- Calendar throws IllegalArgumentException when the hour of the day happens to be skipped due to DST changes.
- Fix will ask Calendar to adjust the time accordingly and get the next closest time
Introduction of a new Transaction API that is more consistent with the style
of Spring's transaction managment. The existing Transaction class was renamed
to TransactionLegacy. All of the non-DAO code in the management server has been
updated to use the new Transaction API.
- code comment turned to javadoc
- added braces to if and for statements to make it look more like the rest
- tests added
Signed-off-by: Laszlo Hornyak <laszlo.hornyak@gmail.com>
- javadoc changed - the old one was copy-pasted from AgentShell
- start and stop method removed - they did the same as the overridden methods
- _counter removed as it was only written, but never read
- remove from _asleep map was moved to a finally block, to make sure it is removed even in case of the thread gets interrupted
- Tests created for the above scenarios.
Signed-off-by: Laszlo Hornyak <laszlo.hornyak@gmail.com>
ACS is now comprised of a hierarchy of spring application contexts.
Each plugin can contribute configuration files to add to an existing
module or create it's own module.
Additionally, for the mgmt server, ACS custom AOP is no longer used
and instead we use Spring AOP to manage interceptors.
- removed methods that were not used
- parseLong, parseInt and parseFloat replaced with the commons-lang NumberUtils call
- Test for the remaining methods
Signed-off-by: Laszlo Hornyak <laszlo.hornyak@gmail.com>
Signed-off-by: Hugo Trippaers <htrippaers@schubergphilis.com>
As we moved to a more Spring managed lifecycle, the lifecycle management
in ComponentContext is not needed. The transition to fully spring managed
will not happen all at once. The management server will be done first and
then AWSAPI and UsageServer will come later. This change allows the
ComponentContext in the context of the management server to be bypassed.
Note, by default, the ComponentContext will run its initialization.
It was implemented by extending the NFS provider. Its validation was updated so that you can pass it a URL containing the
details of a CIFS share. The code that mounts NFS shares was extended to allow it do the same for CIFS shares. Otherwise,
the secondary storage code is left unchanged.
After upgrade the SSVM was staying in connecting/alert state. This was because
while handling the processConnect command for ssvm the management server was
trying to template sync. For resource limit calculation it was trying to get the
remote size of the template. If the template was no longer available a number
format exception was thrown. The process connect wasn't getting completed and
the ssvm used to stay in connecting/alert state. While creating template from
snapshot cloudstack looks for up and enabled ssvms and because there wasn't
any available (as the ssvm was in coonecting/alert state) it used to pick the
wrong resource (LocalNfs*Resource) instead of the NfsSecondaryStorageResource.
Fixed the issue by making sure number format exceptions are avoided so that
SSVM moves to the right state.
No change in behavior if management server has internet connection.
Now if management server doesn't have internet connection download process will not fail. But CS will noly check if the account has already reached or maxed its limits instead of checking if the limit will be breached with the addition of the new storage
Nitin Mehta
Min Chen
miguelaferreira
Hugo Trippaers
Antonio Fornie
wrodrigues
Laszlo Hornyak
Alena Prokharchyk
Devdeep Singh
Daan Hoogland
Sheng Yang
Marcus Sorensen
Kelven Yang
Daan Hoogland
Jayapal
Saksham Srivastava
Alex Huang
Syed Ahmed
Anthony Xu
wilderrodrigues
Kishan Kavala
Darren Shepherd
Sateesh Chodapuneedi
Prachi Damle
Damodar Reddy
Wei Zhou
Darren Shepherd
Donal Lafferty
Prasanna Santhanam
John Burwell
Likitha Shetty
frank
Amogh Vasekar
Edison Su