Changes

Fixes #24136 - Exception when enabling a Monitoring Service
Since OpenJDK 9, attaching of an agent to the current JVM has to be enabled by a system property.
See https://bugs.openjdk.org/browse/JDK-8180425
Signed-off-by:Ondro Mihalyi <mihalyi@omnifish.ee>
 (commit: dd06c97)

Reactivate old CoyoteAdapter normalisation

Signed-off-by: Andrew Pielage <pandrex247@hotmail.com>
 (commit: f8f2d48)

Use exceptions

Signed-off-by: Andrew Pielage <andrew.pielage@payara.fish>
 (commit: 3731928)

FISH-6775 Adjust javadoc

Signed-off-by: Andrew Pielage <andrew.pielage@payara.fish>
 (commit: 4105e6b)

Cherry-pick fixes

Signed-off-by: Andrew Pielage <andrew.pielage@payara.fish>
 (commit: 069c575)

Checkstyle violation

Signed-off-by: Andrew Pielage <andrew.pielage@payara.fish>
 (commit: 7b5b301)

Errant space

Signed-off-by: Andrew Pielage <andrew.pielage@payara.fish>
 (commit: 1da836d)

FISH-6775 Don't use exceptions

Signed-off-by: Andrew Pielage <pandrex247@hotmail.com>
 (commit: 76864a3)

FISH-6775 Fix detail message

Signed-off-by: Andrew Pielage <pandrex247@hotmail.com>
 (commit: da6122b)

Don't use user.home in tests

- better use temp directory.
- in this case now we can be sure that such directory is not a random match.

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: 40b9f69)

Don't use user.home as a default output directory

- I believe that this is rather some internal feature and I wasn't able to
  find any usage. Maybe it should be rather deleted. Why?
  1) Because it generates some file to a directory, but I could not find it
     in any log or documentation.
  2) Because it uses rest api, so it is a server resource, generating file
     on server system. How is such remote file useful to the admin client?
     It should be documented at least.
- So for now I replaced the default directory - if it generates something,
  client must know where. If the generated output is for server, it should
  be generated in appserver subdirectory, not in user's home dir.
- Or I missed something.... not documented.

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: dab6fb0)

Removed unused field initialization

- these paths are set by PECoyoteConnector
- another user.home usages, quite misleading

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: 380aa49)

Deprecated getDomainRoot removed

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: c9c98f2)

Using org.glassfish.hk2.api.PopulatorPostProcessor instead of deprecated copy

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: 764c431)

Renamed ServerPropertyITest

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: e95a390)

Fixed StatusGenerator + test

- the fix is not complete, but at least it doesn't crash with NPE

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: 7197b90)

AdminCache - trywith, logging, synchronization

- however synchronization here still doesn't guarantee anything, usages
  are not atomic

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: a7e8523)

Removed LINUX_HACK constant

- Things like this should not be in production code.

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: d5e0282)

Fixed asadmin logging - ignored exceptions

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: f7922fb)

CliConstants revisited - new env properties and values

- AS_STOP_TIMEOUT to override default timeout for the stop domain command.
- AS_START_TIMEOUT to override default timeout for the start domain command.
- RESTART_CHECK_INTERVAL_MSEC reduced from 300 to 10, performance and timing
  are completely different than before 20 years.

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: f848c94)

Directory for GF client can be set by the GF_CLIENT_DIR env property

- by default it is still in ${user.home}/.gfclient

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: 6639484)

Fixed server restart - more robust, added validations

- Added host name validation with timeout
  - 5 seconds is an extreme value for the local server
- Honoring command-line arguments so they are not overwritten from domain.xml
  which could be updated since startup and doesn't contain correct values
  usable to stop the server.
- WARN: ServerDirs.serverName doesn't contain the server name, but instance
  name or domain name. Probably terminology yet from SunONE.
- Replaced JPS and JSTACK usages by ProcessHandle usages
- WARN: restart-domain leaves zombies on some systems
  - visible on Jenkins TCK
  - these processes are marked as <defunct> and operating system decides when
    they will be cleaned up. It probably relates to Jenkins which reads IO
    from the container, so stese streams remain always open.
- Fixed NPE in ClassPathUtils
- Added AfterAll to ClusterITest
- Added AS_TRACE support to admin-tests - bound to FINEST in tests
- UnknownHostException is now an error for asadmin, it doesn't mean that
  the server is stopped.
- Command line --host has higher priority than domain.xml and default value
  - However it is not always true, ie restart-domain must use the arg value
    for stop, but domain.xml value for start.
- Generalized waitFor, added time measurement.
- Using addSuppressed to keep all errors (visible with AS_TRACE)
- Fixed situation when there was no asadmin truststore - load null and use empty
- JavaClassRunner revisited
  - Avoid "oversanitization", use the same value as the last time
- ServerDirs revisited (quickly)
  - Added getter getLastPidFile and javadoc explaining the difference
  - pid file uses pid of running server, is deleted when server stops
  - pid.prev file uses pid of last started server, is not deleted, just overwritten
- Using Thread.yield when waiting rather than sleep
  - More assertive, even aggressive
  - Increases probablity of detection of misbehavior if there's any
  - Faster, we don't wait if not necessary

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: eabb6ff)

IMQ sometimes stops logging after the server

- maybe server should wait until IMQ stops, we will take a look later

Signed-off-by: David Matějček <david.matejcek@omnifish.ee>
 (commit: 8102a59)