public class Prms extends BasePrms
To add hydra framework parameters, simply add a field definition and javadoc
anywhere in the source for this class (ideally among related parameters), like
so: public static Long someNewHydraParameter;
| Modifier and Type | Field and Description |
|---|---|
static Long |
alwaysDoEndTasks
(boolean)
True if ENDTASKs should be run before stopping the test, regardless of
whether previous tasks had errors.
|
static Long |
archiveStatsForHostAgent
(boolean)
Whether to archive statistics in hostagents.
|
static Long |
checkTaskMethodsExist
(boolean) If true (default), the hydra parser will verify that task
methods are on the master classpath.
|
static Long |
clientShutdownHook
(String pair)
Shutdown hook to invoke in the event of test failure.
|
static String |
DEFAULT_ARCHIVER_ACTION |
static int |
DEFAULT_SLEEP_BEFORE_STARTING_HOSTAGENT_SEC |
static Long |
derbyServerClassPath |
static String |
DISABLE_CREATE_BUCKET_RANDOMNESS_PROPERTY |
static Long |
disableCreateBucketRandomness
(boolean:false)
Guarantees that both buckets and primary buckets will be perfectly
balanced when buckets are pre-created using the PartitionRegionHelper
assignBucketsToPartitions(region) method.
|
static Long |
doInitTasksSequentially
Deprecated.
as of GemFire5.1, use the "sequential" task attribute
instead, which allows individual INITTASKs and CLOSETASKs
to be designated sequential.
|
static Long |
doStartAndEndTasksLockStep
(boolean)
True if STARTTASKs and ENDTASKs should run in lock step, so that each
task is completed by all appropriate clients before scheduling the
next task.
|
static Long |
dumpStacksOnError
(boolean)
Indicates whether to dump stacks on all processes upon a test error
other than a hang.
|
static Long |
errorOnFailedGFStartup
(boolean)
True (default) if hydra should log a failed hydra run if a GemFire
system fails to start.
|
static Long |
extraDerbyServerVMArgs
(String)
The extra VM arguments, if any, to pass to the derby server startup VM.
|
static Long |
finalClientSleepSec
(int)
The time, in seconds, for client threads to pause before shutting
down when requested to by the master controller, e.g, to allow time
for events to arrive and be processed.
|
static Long |
haltIfBadResult
If true, the hydra master controller will stop distributing test tasks as
soon as a client thread reports an error.
|
static Long |
initialClientSleepSec
(int)
The time, in seconds, for client threads to pause before registering
with the master controller, e.g, to make it possible to attach a
debugger before the action starts.
|
static Long |
jmapHeapDumpOptions
(String)
Heap dump options to pass to jmap in the heapdumprun script generated
by hydra.
|
static String |
LEAVE_ARCHIVERS |
static Long |
logLevelForHostAgent
(String)
The hostagent log level.
|
static Long |
manageDerbyServer
(boolean)
Indicates whether master should manage the Derby server.
|
static Long |
manageLocatorAgents
(boolean)
Indicates whether master should manage GemFire distribution locator
agents.
|
static String |
MASTER_HOST_PROPERTY |
static String |
MASTER_PID_PROPERTY |
static Long |
maxClientShutdownWaitSec
(int)
The maximum time, in seconds, hydra should wait for any client vm to
shut down.
|
static Long |
maxClientStartupWaitSec
(int)
The maximum time, in seconds, hydra should wait for any client vm to
start up and register its clients with the master controller.
|
static Long |
maxCloseTaskResultWaitSec
(long)
The maximum time, in seconds, hydra should wait for a thread executing
a CLOSETASK task to return a result.
|
static Long |
maxEndTaskResultWaitSec
(long)
The maximum time, in seconds, hydra should wait for a thread executing
an ENDTASK task to return a result.
|
static Long |
maxHostAgentShutdownWaitSec
(int)
The maximum time, in seconds, hydra should wait for any hostagent to
shut down.
|
static Long |
maxHostAgentStartupWaitSec
(int)
The maximum time, in seconds, hydra should wait for any hostagent to
start up and register with the master controller.
|
static Long |
maxResultWaitSec
(long)
The maximum time, in seconds, hydra should wait for a thread executing
a task to return a result.
|
static Long |
randomSeed
Hydra relies on a random number generator for purposes such as task
scheduling and getting values for "range" and "oneof" parameters.
|
static Long |
removeDiskFilesAfterTest
(boolean(s))
Tells hydra whether to remove disk files after a test, if they exist.
|
static Long |
roundRobin
(boolean)
True if TASKs should be scheduled sequentially in a round robin fashion
(with respect to the logical hydra client thread IDs).
|
static Long |
serialExecution
(boolean)
True if TASKs should be scheduled sequentially in a random fashion.
|
static Long |
sleepBeforeShutdownSec
(int)
How many seconds to sleep before shutting the test down.
|
static Long |
sleepBeforeStartingHostAgentSec
(int)
Number of seconds to sleep before starting each host agent.
|
static String |
SLOW_ARCHIVERS |
static Long |
startLocatorAgentsBeforeTest
(boolean)
Indicates whether to start GemFire distribution locator agents before
the test.
|
static Long |
statisticArchivesAction
(String)
Indicates what action to take with statistic archives when a test hangs.
|
static String |
STOP_ARCHIVERS |
static Long |
stopLocatorAgentsAfterTest
(boolean)
Indicates whether to stop GemFire distribution locator agents after
the test.
|
static Long |
stopSystemsOnError
(boolean)
Indicates whether to stop system resources such as cache servers upon a
test error other than a hang.
|
static Long |
testDescription
(String) Description of the test in this test configuration file.
|
static Long |
testRequirement
(String) Requirement the test is expected to meet.
|
static Long |
testSecurity
(boolean) Indicates whether security is turned on.
|
static Long |
testTopology
(String) Topology used by the test.
|
static Long |
totalTaskTimeSec
Usually, this sets the maximum number of seconds that hydra should remain
in the test task execution loop, giving jobs to clients and evaluating
the results.
|
static String |
USE_IPV6_PROPERTY |
static Long |
useFixedRandomInMaster
(boolean:false)
If true, this parameter causes the hydra master controller to use a newly
minted random number generator seeded with
randomSeed for each
access of a OneOf or Range. |
static Long |
useIPv6
Whether to use IPv6 addresses in master-spawned VMs such as locators
and hydra clients.
|
static Long |
useNFS
Whether the hosts use NFS (defaults to true).
|
static Long |
versionStartAndEndTasks
(boolean)
True if STARTTASKs and ENDTASKs should be versioned using the lowest
version in the version list for each client name.
|
| Constructor and Description |
|---|
Prms() |
| Modifier and Type | Method and Description |
|---|---|
static void |
main(String[] args) |
static String |
statisticArchivesAction() |
dumpKeys, keyForName, nameForKey, setValues, tab, tasktabpublic static final String MASTER_HOST_PROPERTY
public static final String MASTER_PID_PROPERTY
public static final String USE_IPV6_PROPERTY
public static final String DISABLE_CREATE_BUCKET_RANDOMNESS_PROPERTY
public static Long testDescription
public static Long testTopology
public static Long testRequirement
public static Long randomSeed
public static Long useFixedRandomInMaster
randomSeed for each
access of a OneOf or Range. This limits randomization by causing a OneOf
or Range accessed multiple times by master to return a random but constant
value.
This is useful for cases where there are many logical descriptions of a particular type, all of which need to share the same constant, but randomly chosen, value.
Note that if a given parameter uses multiple Ranges or OneOfs, they will all return the same value unless the choices are different.
This setting has no effect on parameters read from hydra clients.
public static Long useIPv6
public static Long disableCreateBucketRandomness
This parameter is a temporary workaround until the product fixes the load balance issue without requiring the system property.
public static Long useNFS
public static Long haltIfBadResult
public static Long checkTaskMethodsExist
public static Long serialExecution
public static Long roundRobin
public static Long doInitTasksSequentially
public static Long doStartAndEndTasksLockStep
public static Long versionStartAndEndTasks
public static Long alwaysDoEndTasks
public static Long manageDerbyServer
public static Long extraDerbyServerVMArgs
public static Long testSecurity
public static Long manageLocatorAgents
public static Long startLocatorAgentsBeforeTest
public static Long stopLocatorAgentsAfterTest
public static Long removeDiskFilesAfterTest
public static Long dumpStacksOnError
public static Long jmapHeapDumpOptions
-dump:live,format=b for 1.6 and
-heap:format=b for 1.5..public static Long stopSystemsOnError
GemFirePrms.stopSystemsAfterTest.public static Long statisticArchivesAction
STOP_ARCHIVERS to shut down the archivers,
SLOW_ARCHIVERS to reduce the sample rate to 60 times its
current value, and LEAVE_ARCHIVERS to leave the archivers
running at their current rate. Defaults to SLOW_ARCHIVERS.public static final String STOP_ARCHIVERS
public static final String SLOW_ARCHIVERS
public static final String LEAVE_ARCHIVERS
public static final String DEFAULT_ARCHIVER_ACTION
public static Long sleepBeforeShutdownSec
public static Long totalTaskTimeSec
maxTaskExecutions is exceeded, hydra
will assume that testing should continue for at least
totalTaskTimeSec. When totalTaskTimeSec is
reached, the master controller will ordinarily wait a bit for late results,
then shut down clients and enter its final reporting and cleanup phase.public static Long sleepBeforeStartingHostAgentSec
DEFAULT_SLEEP_BEFORE_STARTING_HOSTAGENT_SEC. Useful when
starting large numbers of hostagents to avoid timing out from an NFS
overload on the master host.public static final int DEFAULT_SLEEP_BEFORE_STARTING_HOSTAGENT_SEC
public static Long maxHostAgentStartupWaitSec
public static Long maxHostAgentShutdownWaitSec
public static Long logLevelForHostAgent
public static Long archiveStatsForHostAgent
public static Long maxClientStartupWaitSec
public static Long maxClientShutdownWaitSec
public static Long initialClientSleepSec
public static Long finalClientSleepSec
Note that this value should always be smaller than
maxClientShutdownWaitSec.
public static Long maxResultWaitSec
maxEndTaskResultWaitSec.public static Long maxEndTaskResultWaitSec
maxResultWaitSec.public static Long maxCloseTaskResultWaitSec
maxResultWaitSec.public static Long clientShutdownHook
public static void and take no arguments. Defaults to null.
Shutdown hooks are invoked either when the first client reports task failure, as part of client shutdown, or when master detects a hang, in which case the hooks are attempted but not guaranteed (unfulfilled requests are logged at severe level). Shutdown hooks are not invoked on a pass.
Shutdown hooks are invoked concurrently on a single representative thread
in each client VM and are subject to maxClientShutdownWaitSec.
Hydra client threads, including threads with the same logical thread ID,
can still be executing tasks when hooks are invoked. Use extreme care
when using HydraThreadLocals and other state inside shutdown
hook methods.
public static Long errorOnFailedGFStartup
public static Long derbyServerClassPath
Copyright © 2010-2015 Pivotal Software, Inc. All rights reserved.