Friday Apr 11, 2008
I was reading about the persistent
Baital (or vampire) who continually caused difficulties for the legendary king
Vikramaditya with his puzzling tales and I was reminded that I should write the second part to
my earlier blog about recurring problems in GlassFish.
These two installation-related problems pop up really often!
Most Common Problems and Solutions
| Problem |
Solution
|
On Linux or Solaris, GlassFish or Sun Java System Application Server installation fails. Error message says the installer could not create the default domain.
|
Set your environment variable LC_CTYPE to "C".
|
GlassFish or Sun Java System Application Server installer freezes and does not allow you to continue installing after you provide the installation path.
|
This problem occurs because of a previous unclean/partial uninstallation. or failed installation. The solution is to delete Application Server-specific entries from the productregistry file:
* On Windows, the productregistry file is usually at c:\Windows\system32\productregistry. Do NOT delete this file. Search for all application server-specific entries and remove them.
* On Solaris, run /bin/prodreg and uninstall application server.
* On Linux or other non-Solaris Unix systems, the productregistry file is at var/tmp/productregistry. Delete all application server-related information.
|
Wednesday Jan 09, 2008
This blog attempts to summarize what I've learned from forum postings
and personal experiences as I try out various things with
GlassFish 9.1.
Log files
A useful resource for troubleshooting your problems are the log files
(there is a separate
server.log file for each domain, node agent, and
instance). For problems related to starting a domain, nodeagents,
or instance see the server.log.
The location of the
server.log file for domains, nodeagents, and instances:
- domain:
<install-dir>/domains/<domain-name>/logs/server.log
- nodeagent:
<install-dir>/nodeagents/<node-agent-name>/agent/logs/server.log
- instance:
<install-dir>/nodeagents/<node-agent-name>/<instance-name>/logs/server.log
Thread Dumps
Another useful resource is the thread dump. With JDK 5, you
have the very useful jps command. Run the jps command and look for
PELaunch, which denotes a running app server process (node agents,
domains, clusters, etc).
(prompt) jps
If you want to terminate unresponsive app server process use the
kill
-3 <PELaunch PID> command in Unix and
sendsignal
utility for Windows. For more info, see
thread dumps.
Common Problems and Solutions
| Problem |
Solution
|
Cluster instances fail to start.
server.log of the instance shows "connection refused" messages
|
1. Ensure that you followed the
steps here
and here:
2. If you're still having problems, edit your /etc/hosts file and
ensure that your hostname maps to an absolute IP address and not to
127.0.0.1.
|
You cannot log on to the Admin
Console of the DAS or a domain, although it appears that you have have
started it.
|
1. Ensure that you
are NOT using the deprecated startserv command. See the docs
for instructions on how to start a domain.
2. Verify that your hostname is mapped to an absolute IP
address.
3. Ensure that you do not have a firewall running.
4. There is a known issue and a workaround for Mac OS X systems. See
this
forum post.
|
Unable to start the domain.
|
1. If the error messages suggest
that you do not have read-write access to the domain's folder, you
probably installed the application server as a different user and you
are trying to start the domain as a different user.
Ensure that the user who installed the app server provides the app server user with read-write access to app server domain folders.
|
| Although you have installed JDK
5, GlassFish reports that Java 2 SDK version 5.0 or above not found. |
1. Ensure that you are running
the correct version of java to unjar your files. Add
<jdk5_install_dir>/bin to your PATH environment variable. Run
java -version command to ensure that you are running the correct
version.
2. Set JAVA_HOME to your <jdk5_install_dir>. |
Load balancer (working fine
earlier) does not work after restarting the cluster.
|
Ensure that you have created a
health-checker for your cluster and load balancer. Use the create-http-health-checker
command.
|
In-memory replication not
working. Unable to make your web apps highly available.
|
Ensure that you have
enabled/checked the Availability Enabled option while deploying your
application. For step-by-step instructions (either using HADB or
in-memory replication), see this.
|
|
|
|
|
thanks it's a good blog. installation problems wer...