| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
This bug tracks the AT&T bug 355533. The symptom was that drools
pdps that were backing each other up were becoming stuck in a
standby state. The cause was that the election handler thread was
being hung by a call to PolicyEngine.manager.deactivate which shuts
down the topic endpoints when the drools pdp operational state
transitions to disabled. Related problems were that the election
handler heartbeat was NOT blocked when the main thread was blocked
and the IntegrityMonitor forward progress counter was NOT blocked
from incrementing when the election handler thread was blocked.
This prevented the correct failover of the drools pdp to another
healthy one. This change fixes the two causes of the thread blockage,
moves the election handler heartbeat to the main thread and adds an
interface (AllSeemsWell) which is called when the election handler
has stalled/resumed. The AllSeemsWell interface will block forward
progress counter increments when ALLNOTWELL and will resume forward
progress counter increments when ALLSEEMSWELL. In addition, it reduces
the run time of the StandbyStateManagementTest from approximately 8
minutes to approximately 2 minutes. Since this changes classes also
changed by POLICY-444, this change must be merged before POLICY-444
can be merged.
Issue-ID: POLICY-501
Change-Id: I7b8180d11077ccf59b21b6484cb58b5522a3df8f
Signed-off-by: Kevin McKiou <km097d@att.com>
|
|
Under stress, the election handler in feature-active-standby-management
may not get a chance to run within the allowed window causing the
"watcher" thread to kill it and restart it. The run window is
expanded. Also the pdp.updateInterval and pdp.checkInterval
parameter values are increased appropriately in the properties file.
Issue-ID: POLICY-484
Change-Id: I0c1c2d4fdfbc408be929c56e451b92fc7ad4464a
Signed-off-by: Kevin McKiou <km097d@att.com>
|
|
Bump minor version in preparation for Amsterdam
branching.
Change-Id: I7a10894837580ae898573156e47e31f32ae0d041
Issue-ID: CIMAN-120
Signed-off-by: Jessica Wagantall <jwagantall@linuxfoundation.org>
|
|
Released 1.1.1, bump patch to 1.1.2
Issue-ID: POLICY-436
Change-Id: Idc58c1dc0cd66243ca1b94cce2ef4254ef2d87c8
Signed-off-by: Pamela Dragosh <pdragosh@research.att.com>
|
|
Releasing v1.1.0 so we need to update SNAPSHOT
Issue-ID: POLICY-436
Change-Id: I0c24b0c9a5bd67471ad0fb8b8c17b77e5b70b44f
Signed-off-by: Pamela Dragosh <pdragosh@research.att.com>
|
|
Patch 1:
Removed hostPort and added server.TEST properties from
feature-state-management.properties and modified associated
code in IntegrityMonitor and properties in associated JUnit tests.
Patch 2:
Modified in response to comments from Jorge Hernandez. Including
adding contant usage instead of strings and replacing generic exceptions
with specific exceptions.
Patch 3:
Tied property constants to PolicyProperties. Added default property
contants. Added error/warning log statemeents when something other
than expected value is used. Add logging of all constant values
for inspection in debugging.
Issue-ID: POLICY-369
Change-Id: Ie2218b68761e0338642a2ed28ef840b1b6ece1a4
Signed-off-by: Kevin McKiou <km097d@att.com>
|
|
I found multiple issues from latest junit submissions when I built in
my local that I tried to fix.
1. feature-test-transaction - prone to race conditions as we have
experienced in some jenkins builds. There is an assert that checks
if the thread is alive that monitors a policy controller kiesession
sanity. The thread is very short-lived as it will exit right away
since it detects that the underlying "drools session"
does not have an attached rules artifact (brainless).
Removed that check to fix the race condition.
2. With the increment of junits in the PolicyEngine.manager for
multiple packages, it seems that the static instance is reused
across junits (which surprised me), so configuration files that
are not supposed to be read in junits for a package are
read, and for example the lock state could propagate across junits.
I tried to clean all that up to make sure that each junit deals with
what is has created and state does not propagate to other junits.
3. feature-active-standy-management had a missing "junit" dependency.
I generated the effective pom, and indeed did not show, some junits
failed to compile in the test phase. Adding the test dependency
fixed the problem.
As a note, the feature-active-standy-management junits, take over 20 minutes to run.
This time is excessive (see below):
logs$ head -1 debug.log
2017-09-25 21:24:21.630 [main] DEBUG o.o.p.d.c.t.StandbyStateManagementTest.setUpClass(111) - setUpClass: userDir=/media/sf_jh1730/dev/open/LF/git/master/policy/drools-pdp/feature-active-standby-management
logs$ tail -1 debug.log
2017-09-25 21:46:29.801 [Timer-46] DEBUG o.o.p.d.a.DroolsPdpsElectionHandler.run(919) - TimerUpdateClass.run.exit
Change-Id: Ie3167e5f784f35f98fa08997e624c51f976b6501
Issue-ID: POLICY-109
Signed-off-by: Jorge Hernandez <jh1730@att.com>
|
|
Removed unused import
Null pointer exception
Does not evaluate to true
override hashCode
log exception
cast long
public static final
Issue-ID: POLICY-261
Change-Id: I2d90f8503fcc5ed7d13aff31143b8fb69c689e18
Signed-off-by: Pamela Dragosh <pdragosh@research.att.com>
|
|
Patch 1: Adds the active-standby feature to drools-pdp.
This feature provides the state control of the
drools-pdp nodes controlling failover with a site
and across sites. Patch 2: Resolve merge conflict in
packages/install/pom.xml and pom.xml. Patch 3: Resolved
comments from Pamela Dragosh and Jorge Hernandez.
Issue-ID: POLICY-156
Change-Id: I922b3d5d8a464006e9675924bcbc7409d68c08d5
Signed-off-by: Kevin McKiou <km097d@att.com>
|
Kevin McKiou
Jessica Wagantall
Pamela Dragosh
Jorge Hernandez