Age | Commit message (Collapse) | Author | Files | Lines |
|
XACML was released with minor fixes for XML files with comments. Mainly
it contained upgrade to dependencies that had fixed security issues. One
still remains but will not be able to be fixed.
Issue-ID: POLICY-616
Change-Id: Ifaf4f92d8da878a1c49dff5fd028744d01999ff5
Signed-off-by: Pamela Dragosh <pdragosh@research.att.com>
|
|
Unit test expanded for SO POJOs, technical debt removed in eventmanager.
Change-Id: I7045f5af88fd2c15019befe29a0b4fcd69801c49
Signed-off-by: liamfallon <liam.fallon@ericsson.com>
Issue-ID: POLICY-455
Signed-off-by: liamfallon <liam.fallon@ericsson.com>
|
|
Unit test for guard added and technical debt removed.
Issue-ID: POLICY-455
Change-Id: I440766660ddae078013e21b1991ee49c8e488bb3
Signed-off-by: liamfallon <liam.fallon@ericsson.com>
|
|
Mainly making fields private, fixing field and methods to follow Java guidelines, and adding getter and setter methods. Unit test added for all classes in org.onap.policy.controlloop
Issue-ID: POLICY-455
Change-Id: I445b7cfaf9eb921a230bbb72b06ff4455fe003ce
Signed-off-by: Liam Fallon <liam.fallon@ericsson.com>
|
|
This is to prevent potential problems where multiple jar versions are
updated one after the other.
Change-Id: I78835b485e57125efb213cfdefba60351d815916
Issue-ID: POLICY-487
Signed-off-by: Jorge Hernandez <jh1730@att.com>
(cherry picked from commit f2715bc0f272e5e657ed8a82c0f4e520a7c7706d)
|
|
- Renamed modelNameVersionId to modelVersionId in SOModelInfo
- Removed requestID from SO Request body
- As a result of this, an SOResponseWrapper was needed in order to
attach the control loop event requestId to use in the SOResponse
rule in drl (the requestID that is returned in the SO response
is a newly generated requestId with no relation to the Policy
control loop event requestId as per Arthur)
- Updated drl to reflect the above mentioned change
- Updated simulator and junits to reflect changes
- Added an additional null httpResponse check/logger statement in SOManager and RestManager
Issue-ID: POLICY-438
Change-Id: I5f414ba69b60b5565cca9073b47f4c4835e5abbf
Signed-off-by: Hockla, Ali (ah999m) <ah999m@att.com>
|
|
When an operations was denied by guard, it was not writing to the DB at
all. Only in the deny case we start the operation, then
setOperationHasGuardDeny, then finish the operation. Had to add start
operation just before finishing it. This allows the write to occur at
setOperationHasGuardDeny.
Issue-ID: POLICY-345
Change-Id: I665b9cfaa684493e39e731102d4d43ecb39fad7c
Signed-off-by: Temoc Rodriguez <cr056n@att.com>
|
|
The drools engine will now only fire the event rule if the
event is an onset. This prevents abatements from being processed
as onsets.
Issue-Id: POLICY-407
Change-Id: Ic74efa74fbb1dc0e01d7c36a06fdaf48fc8a3238
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
|
|
The drools engine will now send final failure notifications
if there is an error returned from A&AI queries instead of
rejected notifications.
Issue-Id: POLICY-364
Change-Id: Ie9c22686d859c633bad47d223726957cc9599f7c
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
Made sure a final failure is sent and the memroy is cleaned up if we
cannot contruct a vfc request.
Issue-Id: POLICY-350
Change-Id: Ia9bc827572042ae6ff2d2d2851d33034ee86fc49
Signed-off-by: Charles Cole <cc847m@att.com>
|
|
The locking of the target is now taken care of when
the operation manager is initialized. This allows us
to keep the template flow the same and to ensure that
for vFW we will not lock the source if it is not intended
to be the target. NOTE: This will have to be revised in
the future if policy chaining will support having
different targets specified for each policy in the chain.
Additional work was done for hardening the APPC model code
to make sure exceptions are caught and we gracefully end
processing the event with memory clean up and a final failure
notification sent.
Issue-Id: POLICY-367
Change-Id: Ic796d95eb5400067744492f810dd8069ba6241b3
Signed-off-by: Daniel Cruz <dc443y@att.com>
Signed-off-by: daniel <dc443y@att.com>
|
|
generating a control loop project from archetype
(but not from junits) runs into velocity parse failures
as specified in ticket. The existing .drl code
for the line that fails the parsing has been slightly
modified to bypass this problem. In addition a
logging statement has been added to the catch.
Change-Id: I3eec82c9b710cc9d2a13dc68032a8a6074016b40
Issue-ID: POLICY-376
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
Now reject an event if the AAI get query for the event fails. Added test
cases for this behavior.
Fixed some issues from my previous pull request.
Issue-Id: POLICY-314
Change-Id: I674e95302a271423c307a88d061064e12e491a7a
Signed-off-by: Charles Cole <cc847m@att.com>
|
|
Errors from AAI after a Named query now throw an AAIEXception. This is
caught in the template to allow the resources to be removed from memory
and a final failure to be thrown.
Issue-ID: POLICY-314
Change-Id: I319d29ef537b2d01ca288622aac1d9dbbe05f5eb
Signed-off-by: Charles Cole <cc847m@att.com>
|
|
This fix enables the pdp to stop events with
different requestId's/controlLoopNames from
being processed if the target entity specified is
already being processed by another event/controlloop.
The lock timeout rule is removed to stop the event from
waiting to obtain a lock if it was not already required.
Handling of the lock denied case is taken care of in the
event manager rule instead of in a seperate rule to avoid
further unneccessary computation.
A new control loop test case was created that is for testing
general control loop errors. The test case included is for
testing that the lock will deny a request with a duplicate
target entity.
Issue-ID: POLICY-341
Change-Id: Ib62286eff74ec22b2c645f32f385cb331fb1dff3
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
|
|
- Removed nonBaseIndex since there is only base vf Module in v11 response
- Reverted to checkExtractedFields()
- Modified the AAI Simulator to return the v11 response
- If AAI NQ fails then the SO Request will not be constructed. In this
case, we are retracting objects from the transaction and inserting policyResult.FAILURE into current operation to
prevent null values in finishOperation
Issue-ID: POLICY-339
Change-Id: I1c48fac4f9a51467ad54b0706543f5d1baa70389
Signed-off-by: Hockla, Ali (ah999m) <ah999m@att.com>
|
|
Missing abatement=false for Use Cases other than vCPE. Added DCAE
Control Loop Event message for ABATED.
Issue-ID: POLICY-346
Change-Id: Ic8890dd81607ff8657a6f9e49cd9be51cb410fe1
Signed-off-by: Pamela Dragosh <pdragosh@research.att.com>
|
|
PAP currently generates a set up rule that inserts a Params object
when the session is loaded. Since the official template also
has an equivalent rule that is used by the drools applications
team to perform initializations tasks, at runtime we end up with
2 Params objects with the same control loop data. This has
sideeffects where the same rule becomes true for 2 Params objects
therefore performing duplicated actions which introduces somewhat
anomalous behavior.
The PapParams is introduced to satisfy current PAP needs for
management of rules. The structure is included here only for
those purposes but it carries no significance at runtime during
drools executions since the rules go by Params instead of PapParams.
In the near future, the plan is to have the PAP not generating
SET UP rules.
There is a companion review on the PAP side to refer to PapParams
instead of Params to fulfill the resolution of POLICY-302.
Change-Id: I2f9ed3cee3d5d5209eecd43823cfc44a4b937600
Issue-ID: POLICY-302
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
The existing APPC Response Cleanup Rules will drop responses
that are not specific for that control loop. For example
vDNS control loop will drop LCM Response for vCPE control loops
preventing progress of the use cases.
This only happens when the control loops coexist with the
current set of rules.
Change-Id: I5a9ae26e39ecce8dd7f1098d02b6c6b97650b67a
Issue-ID: POLICY-298
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
Change-Id: I05c577e8760f5b4b8e6375f50a327c9dde575e06
Issue-ID: POLICY-260
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
This allows the user to disable or enable guard
through the .properties.environment file. Verified
and tested in a pdp. Properties were added to the
simulators properties file for the simulators to
work out of the box.
Issue-ID: POLICY-259
Change-Id: I0027a5d28f1b30e81bdbe42fa17621b36a61c850
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
This applies changes to make retries work properly
for all control loops. The current design was ignoring
the upper bound of the retries and retrying until either
a success or control loop timeout occured. This is now
fixed to only do retries until the limit is reached that
is specified from the policy.
The operation is now started in the GUARD.PERMITTED rule.
I think this is better because it stops Policy from doing
extra processing if there is a guard deny. This is also
needed so that we can properly do retries for all cases.
The notifications sent in GUARD_NOT_YET_QUERIED and
GUARD.RESPONSE are now more informative with the
message specifying the actor and recipe. The not queried
rule has a message stating that we are sending a query
to guard and the guard response message in the guard
response rule specifices the result from guard.
During a retest of vDNS it appeared that the archetype
template was no longer working, this was because there
were changes in the JUnit template that were not
reflected in the archetype template. These were added
to archetype and vDNS is verified to work again.
Another small fix needed was making sure the action for
vCPE is "Restart" instead of "restart". APPC will
reject our request if "Restart" is not sent as the
action.
Issue-ID: POLICY-259
Change-Id: I28dd3c9a629d297b408775a01afadd5c19351e37
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
Added guard authentication, client authentication, and environment http
headers to the pdp-x guard restful request. Properties are set to
PolicyEngine.manager. Property getter, setter, and properties in
guard/.../Util.java. Test properties defined in junits themselves. Added
code from GuardContext.java to grab the properties and make restful request.
PolicyGuardXacmlHelper now closely resembles GuardContext in order to mimic
functionality. Guard url no longer is passed into CallGuardTask, it is now
a property.
Issue-ID: POLICY-260
Change-Id: I5b144764828b6da0e7b738a578e4f6596a0f4f36
Signed-off-by: Temoc Rodriguez <cr056n@att.com>
|
|
This is a WIP for getting guard to work. Currently
the A&AI GET and named queries are connecting to
the simulator and working in the labs. Guard is not
connecting to simulator and needs further analysis.
Issue-ID: POLICY-259
Change-Id: If9875bfd83cbd82dcae04a876b3818ec9c07b1f7
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
This includes VoLTE support for the amsterdam template.
As of right now all use cases are able to get a final
success with guard disabled and no A&AI GET queries.
Currently, the GET query is not working in the lab.
I will address this in another pull request.
Issue-ID: POLICY-259
Change-Id: I216aa84daaa1e93ab12c56ee53283b4cb413a7c3
Signed-off-by: Daniel Cruz<dc443y@att.com>
|
|
These changes are necessary to get the use cases working
on an actual PDP. So far vCPE, vFW, and vDNS have been tested
successfully on a pdp. These are priliminary tests to just
test a FINAL SUCCESS without A&AI GET queries being used.
A&AI named queries for vFW and vDNS have been verified to
work.
Changes in the template were made for SO as there were
changes in the JUnit template that did not get reflected
in the archetype template.
Changes in the code were made in various places to use
the environment properties for REST urls, usernames, and
passwords. Due to VoLTE not being tested yet, the url
for VFC is still hard coded and will be changed later.
It has been confirmed with APPC that LCM will have two
topics: APPC-LCM-READ and APPC-LCM-WRITE. Policy needs
to sink to APPC-LCM-READ and pull from source
APPC-LCM-WRITE. This has been reflected in the controller
properties file.
Issue-ID: POLICY-259
Change-Id: Ib9a8df07ae5ad9d3052c88907c1e522952af474d
Signed-off-by: Daniel Cruz <dc443y@att.com>
|
|
|
|
Removed the embedded guard decision and replace with restful call to
xacml pdp to restore guard functionality. Set guard URL with PolicyEngine env properties. Modified templates accordingly.
Issue-Id: POLICY-260
Change-Id: Ic1558a6ebdd5f6d1b74a748f69433f6213dbf984
Signed-off-by: Temoc Rodriguez <cr056n@att.com>
Signed-off-by: Hockla, Ali (ah999m) <ah999m@att.com>
|
|
run successully vcpe case in lab in 2 steps:
1. docker-install load
2. run create-cl-amsterdam (with defaults)
3. injecting dcae.onset and appc.success packaged
with archetypes
Change-Id: Idb20c0078228da962510dbf36dae96aceb43546c
Issue-ID: POLICY-162
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
Change-Id: I3e659d35d3cb397c011657f39ef12b10b0564fc9
Issue-ID: POLICY-162
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
Issue-ID: POLICY-162
Change-Id: Ieb0952379cd854e0ed8e4a3b068b7e29f3b93770
Signed-off-by: Hockla, Ali (ah999m) <ah999m@att.com>
|
|
This is work in progress, the official pom.xml with dependencies,
drl template, and support files for controller deployment
are maintained here. In the near future the junit template
should be consolidated with this one.
Added controlloop.properties.environment, this environment file
will be populated at installation time with the lab's aai url,
etc .. and will be accessible by any drools application such as
control loops through the PolicyEngine interface. Note that PDP-D
server already supports these environment files, so it is just natural.
Therefore, this is the default mechanism to provide to applications, the url, username,
and passwords to use at runtime by the control loops for the time being.
In the future MSB could set them globally here through existing APIs,
or it can be queried by any drools application using MSB library,
doesn't matter.
There's been some trouble playing nicely with the dependencies used
by a control loop application classsloader, and the pdp-d middleware one,
causing issues between dependencies version of libraries. Specifically,
the snakeyaml library does not play well across classloader when using
constructor functionality, note that the snakeyaml libraries are pulled
also from jackson parsers used in the pdp-d. I made a change in ControlLoopProcessor
to specifically tell the "Yaml" object which classloader to use in order to
find the class with the constructor that is intended to be built, otherwise,
yaml libraries use a different classloader that does not have visibility
into the ControlLoopPolicy that is trying to construct, and fails. This also
should respect junits that use the same classloader I pressume and does not
give issues.
Change-Id: I36271d29cdbf8ff861f9c03ff91cf7116927906a
Issue-ID: POLICY-162
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|