Build DataONE Integration Testing Package (Mar 13, 2016 9:10:09 PM)

	Build Artifacts d1_integration-2.0-SNAPSHOT.pom 12.83 KB  view
	Changes Renamed EnvironmentStatusTest to ...TestImpl, so maven doesn't see it as a unit test. (detail) by rnahf switched testSetReplicationStatus_dateModified to use a non-auth MN to test status changes on, just in case this causes problems with replication (this won't fix sysmeta date change issues though) (detail) by andreib Incrementing a few object pids to tests don't use objects that are too old - just in case this is causing a current issue with sysmeta dates seemingly getting updated. (detail) by andreib log the pid and dates that do not match when failures occur. (detail) by leinfelder don't test with cnSandboxUNM1 - seems to be having issues with keep HZ connections alive and passing tests. (detail) by leinfelder getting test errors for .14, try .15 as object suffix? (detail) by leinfelder include write permission for rightsholder (detail) by leinfelder Added public-readable to access policy for created test objects (detail) by andreib Another logging improvement - displaying created auth token (detail) by andreib Fixed a few logging messages (detail) by andreib Added an update test to the MN auth token tests (detail) by andreib Switched auth token tests to using the more complicated ORCID rather than the basic "testId" subject (detail) by andreib Reverted general log level back to WARN, ContextAwareTestCaseDataone logger still at INFO level (detail) by andreib Part of a refactor to query results checking - should've been in cluded in last commit (detail) by andreib Added a query test for auth tokens, added an MN create test using an auth token to create on the MN, not just for authentication/authorization. (detail) by andreib Fixed auth token test for MN - it needed to just get the v2 member nodes (detail) by andreib Adding private key filename property to STAGE test properties (detail) by andreib Changed logging level for d1_integration to INFO, re-added all CNs into token auth test (detail) by andreib Updated dependencies to 2.1.0 snapshot in d1_integration pom (detail) by andreib Added debug logging for auth token created (detail) by andreib No need for skipping CNs since test.dataone.org key should work for all of them. (detail) by andreib Leaving the public cert property null (rather than empty) just in case (detail) by andreib Updated some failure log messages (detail) by andreib public cert property can actually be null, it's not needed to generate the token (detail) by andreib sandbox test properties public cert needed to point to a different file (not the key file - that contains the private key only). (detail) by andreib Changing the private key and public cert parameters - they need to point to the server's (a CN server's) private/public key  pair. (detail) by andreib Typos in sandbox test properties file (detail) by andreib Added private key and public cert  properties to sandbox test properties file so we can use the auth token tests. (detail) by andreib fixed a couple comment typos (detail) by andreib Added a test that hits echoCredentials with the token. (detail) by andreib Improved some error messages (detail) by andreib Bug fix - the MN call wasn't being done as the CN. (detail) by andreib Modified teh MN isAuthorized test to also procure a test object (get or create) and to do so as the CN (to isolate the isAuthorized part for this test). (detail) by andreib Changed the functionality of the CN auth test to create an object readable by the token subject (AccessPolicy), but created by the CN subject (submitter/rightsHolder). (detail) by andreib Bug fix - incorrect property name in TestSettings fixed (detail) by andreib Modified TestSettings to be able to pick up and set cn.server.privatekey.filename and cn.server.publiccert.filename from test properties files to Settings properties. (detail) by andreib Added a test against MN create/isAuthorized (detail) by andreib Added some logging and cleaned up some imports. (detail) by andreib Added some new classes for auth token tests (mostly placeholders at the moment, untested). (detail) by andreib Added d1_portal to d1_integration pom.xml dependencies so we can use the TokenGenerator for tests. (detail) by andreib Improved a bunch of log/error messages on log tests (detail) by andreib Modified a second log test to skip over MNs on which objects cannot be created. (detail) by andreib Another tweak to resolve test - fixed the logger being used for this class so we can see the INFO messages and removed a break statement causing problems with the logic. (detail) by andreib Performance improvement for last modification to resolve test (detail) by andreib Added code to a resolve api test to iterate through the object list a few more times not only if getSysmeta() fails, but also if the authMN of the object in question is not found in the environment. (detail) by andreib Modified the aggregation test to disregard MNs on which create is disabled. (detail) by andreib Improving on some of the environment tests - removing a couple unnecessary tests for nodes that are registered but down. Fixed some typos. Improved a few log messages. (detail) by andreib Adding some tests that can be run to check on the status of an environment (stuff like which nodes are down, which nodes don't have sync or replication turned on, or to check if sync/repl/services don't match between CN & MN Node documents). (detail) by andreib Small change that should've been committed with the last fix to create a test object instead of procure one. (detail) by andreib Found a subtle race condition in a test and fixed it. (detail) by andreib Need to call v2 /log on nodes that support it and v1 /log on v1 nodes because they only support pidFilter. (detail) by andreib Added an extra error message around creating a log entry (detail) by andreib turned off the v1-only bypass in SANDBOX.  better late than never. (detail) by rnahf added v1-only flag to STAGE properties files (as was done for SANDBOX) (detail) by rnahf adding an ignore to one of the SlowIT tests (testGetLogRecords_eventFiltering) because it was causing severe delays in test environment. (detail) by rnahf Added an extra get() call - there was a bug in metacat and initial create() logs weren't written. Since the test code procures rather than creates the objects, the extra get() should add logs to make the test more resistant. (detail) by andreib Performance improvement for pidFiltering test and added some more comments to clear up what's going on in the test. (detail) by andreib Committed an extra certificate/subject log message for debuggging purposes (detail) by andreib Updated the d1_integration pom.xml file to depend on d1_test_resources version 2.0.0 rather than 1.1.0 (detail) by andreib Added a warning log message if only v1 MNs are available to work with. (detail) by andreib Just removed a variable (that won't change anything) so local cnSubmitter points to the ContextAwareTestCaseDataone variable (detail) by andreib Removed the requirement for v2 MNs from LogAggregationFunctionalTestImplementations - switched to using the highest endpoint version available for the MN where possible, falling back to using the v1 endpoint where it became too expensive to check both capabilities per node available. (detail) by andreib added a property nodelist.contains.v2.mn that to set the nodeListContainsV2Mn property.  Set the property to false in the context.SANDBOX.test.properties. Adjusted TestSettings to allow the property to be set from the command line (-D arguments) (detail) by rnahf Added checks in a few functional test implementations for the ContextAwareTestCaseDataone property that is set based on whether we have v2 nodes in the environment. Some tests will be skipped without v2 nodes in environment. (detail) by andreib Added a nodeListContainsV2Mn property. It was being set automatically based on the CN node list, but disabled that for now per a discussion with rob since we may just change it to be an environment property. (detail) by andreib added optimization to the createTestObject method in SidCNTestImpl to minimize wait-to-sync time. (detail) by rnahf Added a more specific error message and modified some v2 node checks to be even more strict (detail) by andreib Reverted recent change - object creation test code, in the case that we have a v2 CN.create() with no v2 MNs in the environment, will not fall back to a v1 MN for the authMN because this changes the expected behavior of some tests (tests that are expecting a v2-specific object to be created, aka an object with a v2 authMN). Better that it fail for not finding a v2 authMN than sneakily doing a v1 create where a v2 create is expected. (detail) by andreib Object creation test code fix - in the case that we have a v2 CN.create() with no v2 MNs in the environment, fall back to a v1 MN for the authMN. (detail) by andreib Improved a few log messages. (detail) by andreib Removed a line that shouldn't have been part of last commit (detail) by andreib Fixed a bad typo (detail) by andreib changed dataone.it.cnode.submitter.cn to ORC, due to taking UNM Sandbox CN out of service. (detail) by rnahf UNM cn has been temporarily removed from Sandbox environment (no longer in node list) and we can't use the UNM cn cert to create objects. Switching to the UCSB cert. (detail) by andreib Added some error logging (detail) by andreib Added a warning (troubleshooting step if required) to the d1_integration README. Also fixed some numbering. (detail) by andreib Web Tester fix - web test annotations for test labels and test descriptions were not being applied in many cases. The problem was just in cases with multiple @WebTestImplementation classes; iterating over the possible implementation classes and checking for the test method should've continued looping on failure to find said method - as it could've been located in the next implementation class. (detail) by andreib Improved some output for web tester with slightly (very slightly) less developery ignore comments (detail) by andreib Fixed an edge case that could mess up a replica being fetched if we ever ended up with a v2 MN that does not support the v1 API. (detail) by andreib Added a few annotations for the webtester's documentation (detail) by andreib Messages improved inside of some exceptions thrown from replica fetching, also bumped up the maximum replication wait in the CN synchronize test (won't affect test run time if replication is no over-burdened) (detail) by andreib Minor log message fix (detail) by andreib Made a change to the RetryHandler's last iteration: we don't want to throw a TryAgainException outside of the RetryHandler, so attempt to throw the exception encapsulated by the TryAgainException if there is one (detail) by andreib Bumped up the maximum replication time yet again since it sometimes takes forever in practice (if replica for pid is resubmitted) (detail) by andreib Performance fix - decreased the desired number of replicas in a couple tests to the number of valid targets minus the origin MN, so CN won't need to re-attempt any replication hopelessly (detail) by andreib Bumped up the maximum replication time yet again since it took just over 40 minutes in practice (if replica for pid is resubmitted) (detail) by andreib Added a sanity check to testUpdateSystemMetadata_CNCertNonAuthMN (detail) by andreib Added a more careful non-auth MN check to a test, it may have been failing in some cases. (detail) by andreib Minor log message change (detail) by andreib Added a few log messages, fixed another log message, added a slight performance improvement. (detail) by andreib Slight performance improvement and logic fix to the sysmeta sync check (detail) by andreib Bumped up the maximum replication wait time in another test (affects worst-case scenario only since we poll sysmeta) (detail) by andreib Bumped up the maximum replication wait time in a test (affects worst-case scenario only since we poll sysmeta) (detail) by andreib Log message fix (detail) by andreib Improved logging in testUpdateSystemMetadata_CNCertNonAuthMN and bumped up the max replication wait time a few minutes. (detail) by andreib Fixed a couple typos from last commit (detail) by andreib Rewrote part of a log access test to make it less susceptible to read time-outs - split up cn subject calls (individual pid filters) rather than doing the paged getLogRecords. (detail) by andreib Performance improvement - Completely refactored V1/V2 interop tests to create all objects, then test for sync/replication before running the tests. This way we wait the maximum time once (worst case) and not for each created object. (detail) by andreib Made replica fetching a bit more specific in the setReplicationStatus test case, switched possible MNs for updateReplicationStatus replica check to v2 MNs (shouldn't affect test, just save a bit of time) (detail) by andreib Modified another replica check on the test modifiying replica status to remove the necessity for a completed state. (detail) by andreib Modified the replica check on the test modifiying replica statuses to remove the necessity for a completed state. (detail) by andreib Improved error message (detail) by andreib Modified an interface resolution call in CommonCallAdapter (detail) by andreib Performance improvements - short circuiting a handful of extra checks. (detail) by andreib Performance improvements (switch to polling up to a total wait time rather than full wait time) and added some stricter replica checks to the polling. (detail) by andreib Improved some performance - for object creation onto an MN with expected sync to CN, switched to polling for the CN sysmeta on intervals instead of waiting (detail) by andreib Polling cn.getSysmeta for sysmeta with replicas will retry if no MN replicas are found (detail) by andreib Polling cn.getSysmeta for sysmeta with replicas will retry if no v2 MN replicas are found (detail) by andreib Increased the max wait time for sync / replication to happen, since we're now polling for the updated sysmeta, it won't increase the run time of the tests except when CN sync is under heavy load and sync is trying to catch up anyway (detail) by andreib Split v1/v2 interoperability tests' sync and replication checks into separate steps to get better error reports. Also improved performance of tests by using polling approach and applied it to those steps separately. (detail) by andreib fixed 2 typos (detail) by andreib Updated SystemMetadataFunctionalTestImplementation, added rob's RetryHandler to rest of methods that need to wait on sync/replication, modified some assertions and logging (detail) by andreib added Identifier.getValue() to logging and assert messages.  Created a RetryHandler class to avoid compulsory 10 and 5 minute waits on synchronization. (detail) by rnahf Reverting an accidental debugging change (detail) by andreib Some more logging improvements (detail) by andreib Added a bit of extra logging (detail) by andreib Tweaked number of replica targets required (detail) by andreib Bug fix - added a catch block so we can skip nodes whose getCapabilities endpoint is down (detail) by andreib Added some more logging to testUpdateSystemMetadata_CNCertNonAuthMN (detail) by andreib Removed an out-of-date assertion (detail) by andreib Added some more logging to testUpdateSystemMetadata_CNCertNonAuthMN and removed v1 nodes as replica targets (I verified that it was in fact an error that we were replicating to those) (detail) by andreib Bug fix to replication wait time for testUpdateSystemMetadata_CNCertNonAuthMN (detail) by andreib Bug fix to replication policy and replica locating in testUpdateSystemMetadata_CNCertNonAuthMN (detail) by andreib Changed test object update code to make it easier to debug, made public readable (detail) by andreib A couple typos in logging (detail) by andreib Only MN admin and CN should be able to access an MN's log records, the object's rightsHolder currently cannot access its logs; modified the tests accordingly. (detail) by andreib Removed some out of date assertions (detail) by andreib refactored some logging statements in createTestObject to remove printStackTrace commands. (detail) by rnahf roll back change about other id logging - possible NPE (detail) by leinfelder some logging improvements (detail) by andreib reverted a change to sync wait time that shouldn't have been committed (detail) by andreib a bit more logging (detail) by andreib logging addition (detail) by andreib removed errant use of org.hsqldb.lib.StringUtil in a unit test case. (detail) by rnahf typo fixed (detail) by andreib Using a new object on each run of a deleteReplicaMetadata test now (detail) by andreib add other identifier to test failure message. (detail) by leinfelder Improved some error logs. (detail) by andreib Added an assertion on the number of version 2 MNs available (detail) by andreib Bug fix to recently added test - checking for replicas on v1 and v2 MNs (detail) by andreib Fixing a possible NPE (detail) by andreib Fixed a bug in procureTestObject, had introduced a possible validation exception in recent refactor. (detail) by andreib Fixed a problem with the content of ContextAwareTestCaseDataone.getCoordinatingNodeIterator(). When using context.label, it would previously return a list of all CNs found through listNodes(), including the round robin CN. It now tries to filter this node out if the CN list size is >1 (so we don't risk an empty CN list). (detail) by andreib Modified SystemMetadataFunctionalTestImplementation tests to assert that sysmeta.dateSysMetadataModified is not changed as a result of the MN.updateSystemMetadata call (and therefore the resulting CN.updateSystemMetadata call) (detail) by andreib Fixed a typo (detail) by andreib use same subject as creator of  OREs when retrieving? (detail) by leinfelder Commented out a few tests against MN.query() for functionality that I found out is not supported yet - may put them back in after further discussion, if it's something we decide we want. (detail) by andreib Bug fix for a MNSystemMetadataMutabilityImplementations test - had out of date serialVersion (detail) by andreib Modified ContextAwareTestCaseDataone.procureTestObject to also work for a given ReplicationPolicy. Changed the test methods of MNSystemMetadataMutabilityImplementations to get-or-create a test object rather than just create one so it's less susceptible to failing or unavailable services in the environment - and should pick up the test on subsequent runs. (detail) by andreib More logging improvements (detail) by andreib Made change to libclient for getLogRecords - pidFilter changed to idFilter per documentation #refs 7432 (detail) by andreib print stack trace for resolve issue. (detail) by leinfelder Committed some functional tests that check for modification of sysmeta's modification date during replica-related CN calls (detail) by andreib do not add original pid with an obsoletedBy property (I believe this is not allowed) (detail) by leinfelder Bug fix for a CN getLogRecords test (detail) by andreib try shorter nodeId - might be as simple as that to fix test case. (detail) by leinfelder refs: #7419 #7426.  removed IdentifierNotUnique throws from hasReservation definitions and implementations (libclient, sync, cn_rest, d1_integration). (detail) by rnahf More logging improvements (detail) by andreib More logging improvements (detail) by andreib More logging improvements (detail) by andreib More logging improvements (detail) by andreib Improved logging (detail) by andreib Added some sanity checks, made a couple bug fixes to a query test (solr really doesn't like ':'s), re-enabled the aggregation test (detail) by andreib A logging improvement to log and query functional tests (detail) by andreib A couple additions and fixes to log and query functional tests (detail) by andreib Filling out one of the new log tests further (detail) by andreib Added to log and query tests, making a handful of logic fixes, adding logging, adding another test, etc (detail) by andreib Overloaded an object creation method in ContextAwareTestCaseDataone (detail) by andreib Added tests for the log and query endpoints to test access and to test aggregation to CN from MNs (that one however will require modification to the log aggregation schedule to be run). Also made a small logging fix to v1/v2 interop tests. (detail) by andreib restoring d1_integration (detail) by vieglais Modified an updateSystemMetadata test - added a search for valid replica targets and made the replication policy more specific to what's available (detail) by andreib try null for getting the session for verification-not-auth test. (detail) by leinfelder Small improvement in logging, split up a few test setup calls in testUpdateSystemMetadata_CNCertNonAuthMN (detail) by andreib compilation failure (detail) by leinfelder include verification test using cn subject (detail) by leinfelder More logging changes (detail) by andreib Some minor logging improvements (detail) by andreib For easier debugging, changed the created objects in sysmeta synchronize/replication test to be publicly readable (detail) by andreib fix compilation errors from testArchive()  removal (detail) by leinfelder Another bug fix to the v1/v2 interop query test (detail) by andreib Also removed the archive() sid test in the common sid test class as it no longer applies (detail) by andreib Bug fix to a v1/v2 interop query test, to parsing number of results (detail) by andreib Modified the V2 test definitions classes to match the removal of the sid-related tests from the CN sid tests. (detail) by andreib Bug fix - v1 call adapter creation (detail) by andreib Modified the v1/v2 interop tests against v1 sysmeta methods (like archive, setRightsHolder, etc) to hit both the v1 and v2 endpoints and check for NotAuthorized, and removed the corresponding methods from the sid CN tests (since we're expecting exceptions and there's no longer a need to check valid return results) (detail) by andreib Fixed a sneaky bug that was preventing the view call from succeeding for MNs. (detail) by andreib Modified CN sid tests that deal with modifying parts of the system metadata to now expect NotAuthorized exceptions - also added back in the test for the methods setReplicationPolicy - whether it can take a sid or not, the method is only usable against v2 objects (may be redundant with another test, but no harm) (detail) by andreib Modified CN sid tests that deal with modifying parts of the system metadata to now expect NotAuthorized exceptions - the methods archive, setRightsHolder, and setAccessPolicy say they can take sid parameters in the documentation (this will need to be updated because of a recent decision to make these methods only usable against v2 objects - meaning no sid exists for them) (detail) by andreib One of the old tests was attempting a CN.setAccessPolicy on a v2 object; created a separate test that expects failure and switched it out. (detail) by andreib One of the old tests was attempting a CN.setRightsHolder on a v2 object; created a separate test that expects failure and switched it out. (detail) by andreib Logging some extra warnings and modified a setup step to fix a possible node conversion data loss issue. (detail) by andreib We need to be able to get a valid MN reference to use as an authoritative MN on test objects created; added code to ContextAwareTestCaseDataone to do this for calling CNs as well as MNs. (detail) by andreib @ignoring a unit test for the pooling connection manager that should not have been a regular test. (detail) by rnahf use cnSubmitter and cnDevUNM2 for setting up tests (detail) by leinfelder more logging to failure message (detail) by leinfelder Added more specific logging to systemMetadataChanged test method utilizing the multiple environments' CN certs (detail) by andreib log status before testing (failing). mn.dirtysysmeta test. (detail) by leinfelder Increased wait times for sync and replication since MN object creation succeeds sysmeta fetching from CN is failing every time (detail) by andreib Some improved logging (detail) by andreib include cnDevUNM2 (certificate) when testing MN.systemMetadataChanged. Hopefully will allow it to succeed in DEV2 testing environment. (detail) by leinfelder Some small fixes to resetting the node capabilities on finishing sysymeta sync tests (detail) by andreib Found a bug in package creation code that created invalid sysmeta (detail) by andreib Had to update the v1 cn calls that update system metadata to use the fetched metadata's current serial version due to the recent implementation change (detail) by andreib Bug fix to test package setup method (detail) by andreib Removed a second assertion from updateSystemMetadata test using CN cert - there's no rule about serialVersion or dateSystemMetadataModified changing on the CN.updateSystemMetadata call (happens CN REST to CNStorage). (detail) by andreib Fixed a bug in test: CNUpdateSystemMetadataTestImplementations.testUpdateSystemMetadata_NotAuthorized_RightsHolder - call to updateSystemMetadata wasn't actually using the rightsHolder cert. (detail) by andreib CN.updateSystemMetadata null serialVersion test also modified to not expect exception (detail) by andreib Changed an updateSystemMetadata test (using cn cert) to compare with the date in the previous sysmeta version rather than to that from a ping call. (detail) by andreib CN.updateSystemMetadata null serialVersion test also modified to not expect exception (detail) by andreib Fixed error message for the added sysmeta with no serialVersion test (detail) by andreib Reverted change to package creation - created with rightsHolder subject (detail) by andreib Added some logging (detail) by andreib Removed a handful of lines in integration tests that were setting the date modified on system metadata - this is now managed solely by the CN. (detail) by andreib Changed the MN test for updateSystemMetadata with a null serialVersion to no longer expect an exception. Added corresponding test in implementations class (not replacing old one since that still applies to the CN test). (detail) by andreib Updated interop tests dealing with objects that have a v2 authMN but try to use a v1 CN metadata altering call - these should fail with a NotAuthorized per the last maintenance discussion. (detail) by andreib Refactored MNUpdateSystemMetadataIT method to make use of existing code, removed a now-false assertion (to match sysmeta change to handling serialVersion), removed a setting of the dateSysMetadataModified to match same changes in implementation. (detail) by andreib exclude older commons-io from being pulled in with solrj. hopefully this allows bagit library to use newer methods. (detail) by leinfelder test for same content in bagit manifest listing, not same checksum since the manifest order is not guaranteed. https://redmine.dataone.org/issues/7394 (detail) by leinfelder Switched created test packages to be public readable for testing purposes (detail) by andreib Removed some commented-out code (detail) by andreib Minor bug fix to a CNCore test (detail) by andreib Reverted a change I had made for testing purposes. (detail) by andreib Modified the MN API test classes to work with a context.label specifying MNs of both v1 and v2 - made sure that if multiple MNs are given by the member node iterator, then v1 nodes are excluded and only v2 MNs are tested (since these are v2-only methods). (detail) by andreib Modified the view and package functional test classes - made sure that if multiple MNs are given by the member node iterator, then v1 nodes are excluded and only v2 MNs are tested (since these are v2-only methods). (detail) by andreib Modified the view and package functional test classes - made sure that if multiple MNs are given by the member node iterator, then v1 nodes are excluded and only v2 MNs are tested (since these are v2-only methods). (detail) by andreib Cosmetic fix for generated packages - simplifying science metadata pid (detail) by andreib Bug fix in MN sid tests, made sure that if multiple MNs are given by the member node iterator, then v1 nodes are excluded. Also added some more specific error logging. (detail) by andreib Removed check for a replica target on a test in which we're expecting replication to not happen (since it would be v2 to v1 replication). (detail) by andreib Changes in dependency DataONE Java Client Library  ? → (detail) DataONE_Common_Java  ? → (detail) DataONE_JibX_Extensions  ? → (detail)
	Test Result (1 failure / +1) org.dataone.integration.it.functional.v2.CNRegisterFunctionalIT.testUpdateNodeCapabilities