I often get mails from people asking what are the questions generally asked in Weblogic by interviewers.
Generally for an Administrator’s positions, in addition to Weblogic questions, there are a lot of Unix questions and a few questions related to process and case handling.
These are the important questions that I cover while interviewing.
1. How do you differentiate between a server hang and server crash issue?
When a Server crahes, the JAVA process no longer exists. When the Server is hung, it stops responding.
We can use the weblogic.ADMIN utilty to ping the server. In case of a hang situation we can take multiple thread dumps and analyze the cause of hang.
2. What can be the various reasons for a server crash?
a) Native IO
b) SSL Native Libraries
d) Supported Configuration
e) JDBC Driver issue
3. How do you troubleshoot a crash?
JVM crash generates a hs_err_pid file. We need to look into the stack trace of the hs_err_pid file .
If the thread is from a native io, we need to disable native io.
if the stack trace is from the driver, we need to get in touch with the drive team.
Quite possibly its a problem with driver. Changing the type of driver can be a workaround.
If the thread shows it coming from an optimzed code, we can turn of optimization.
If the stack is from native calls of application, its a bug with the application and it has to b modified.
4. Ho do you troubleshoot Server Hang?
We can use java weblogic.Admin PING to check if we get a normal response.
We need to take multiple thread dumps with kill -3 pid on unix and CTLR Break on Windows.
Analyze the thread dump to find the root cause.
5. What can be the reasons of Server hang?
Memory leak, databse query taking a long time to return, Deadlock.
6. What is memory leak?
Memory leak is when objects are not romved from the heap even when they are not required.
7. What are the various causes for OUT OF MEMORY?
a) Insufficient heap size, not able to match the extra load.
b) Objects licing too long, like HTTP Sessions.
c) Memory leak in application code.
d) Full GC not happening due to JVM Bug.
8. How to troubleshoot and overcome such issues?
Gather memory data by enabling GC verbose.
If its due to Http Session, timing out http session after certain interval might help.
Look into the code for jdbc connection handling.
Optimizing the heap size according to the load.
9. When does High CPU Usage occur?
It occurs when one process or one thread utilizes unexpectedly high proportion of CPU.
10. How to troubleshoot it?
In Solaris environment, we need to take pstack and prstack and see what the threads are doing.
In Windows we need to use pslist and process explorer.
11. What is Clustering and what is achieved through it?
Clustering is the grouping together of servers for the purpose of high availability and scalability.
Load balancing and Failover is achieved.
12. How does Cluster Communication Happen?
Members of the Cluster communicate over the Cluster Multicast IP and Port by sending periodic heart beat messages.
13. What is the difference between the Sun JVM and BEA JRockit JVM?
The most well know JVM is the implementation from Sun. The Sun JVM is called HotSpot. The Sun JVM is shipped in the Java Developer’s Kit (JDK) and Java Runtime Environment (JRE) from Sun.
The BEA JRockit JVM from BEA systems is optimized for reliability and performance for server side applications. To achieve this, BEA JRockit JVM uses technologies such as code generation, hot spot detection, code optimization, advanced garbage collection algorithms and tight operating system integration.
14. TUning JVM Parameters.
If you have a single processor, single thread machine then you should use the serial collector (default for some configurations, can be enabled explicitly for with -XX:+UseSerialGC). For multiprocessor machines where your workload is basically CPU bound, use the parallel collector. This is enabled by default if you use the -server flag, or you can enable it explicitly with -XX:+UseParallelGC. If you’d rather keep the GC pauses shorter at the expense of using more total CPU time for GC, and you have more than one CPU, you can use the concurrent collector (-XX:+UseConcMarkSweepGC). Note that the concurrent collector tends to require more RAM allocated to the JVM than the serial or parallel collectors for a given workload because some memory fragmentation can occur.
15. How do you do performance tuning of WLS?
It can be categorized in 4 parts.
a. Application Tuning.
jsp precompilation, ejb pool size cache..
b. OS Tuning
Setting tcp ip parameter.
c. Core Server Tuning.
tune workmanager, tune chuck size and chunck pool size, using performance packs, conenction backlog buffering.
d. JVM Tuning
tuning gc strategy, monitoring garbage collection..