If you are sure* it is the client end then …

Use setTimeout(int timeoutInMilliseconds) in org.apache.axis.client.Stub class you will find in your Service Locator class

* Client end timeouts should usually see the following exception  java.net.SocketTimeoutException: Read timed out … else it may be server side timeout

You will need to append / edit the log4j in the axis2/WEB-INF\classes\log4j.properties file … because axis2 is the web application running using the log4j.properties file in it’s class path.

I hate this because I’d rather not edit a deployed WAR. But, is there a better way?

Step 1 – Make sure mod_deflate is loaded

Make sure the following line is in httpd.conf

LoadModule deflate_module modules/mod_deflate.so

Step 2  - Add the following to u’r httpd.conf

# Method 2: Compress all content, manually excluding specified file types like images
<IfModule mod_deflate.c>
# place filter ‘DEFLATE’ on all outgoing content
SetOutputFilter DEFLATE
# exclude uncompressible content via file type
SetEnvIfNoCase Request_URI \.(?:gif|jpe?g|png|rar|zip)$ no-gzip
<IfModule mod_headers.c>
# properly handle requests coming from behind proxies
Header append Vary User-Agent
</IfModule>
</IfModule>

#3. Restart Apache

You can test this with YSlow … just make sure u remove all browser caching before checking …

1. Remove the jar files from /WEB-INF/lib from your project, so they are no longer a part of the WAR file
2. Copy the jar files to say, {TOMCAT_HOME}\common\lib\myprojlibs
3. edit catalina.properties file in {TOMCAT_HOME}\conf … add to class.loader this path … ${TOMCAT_HOME}\common\lib\myprojlibs\*.jar
4. Remember to add the $ and no ;s etc.

Recently moved one of our products to Amazon AWS cloud (EC2 + EBS + S3) … here are some learnings

• Instances can hang and become unreachable … a lot more than even VPS hosting … you can’t even detach the EBS volume then … so keep a recent Snapshot … so quickly launch a new volume and attach to a new instance
• Any issue … post to Amazon support and the forum … often the forum responds faster …
• Amazon support hours are in US times … even for APAC servers
• Amazon is very powerful, flexible and affordable … but it is less reliable and responsive compared to dedicated servers (comparing an instance to a dedicated server) … so plan accordingly

“A date field shall be of the form 1995-12-31T23:59:59Z The trailing “Z” designates UTC time and is mandatory “

Say your class has a property expiryDate … create another property indexedExpiryDate and index the later field.

So if you want to send data to the index you have to format it …

public Date getIndexedExpiryDate(){

    SimpleDateFormat sdf = new SimpleDateFormat(“yyyy-MM-dd’T'HH:mm:ss.SSS’Z'”);
    Date indexFormatExpiryDate = sdf.parse(sdf.format(getExpiryDate()));
    System.out.println(“indexFormatExpiry: “+indexFormatExpiry);
    setIndexedExpiryDate(indexFormatExpiryDate );
    return indexFormatExpiryDate

}

When you read it back from the index into the bean … it will automatically setIndexedExpiryDate with the date/time local time zone.

How ever, the index itself will hold the date in ZULU time (GMT) … so if you query the index directly you need to adjust for time zone … as queries will not be adjusted for local time zone.

If you are in IST that is 5hrs 30mins ahead of GMT … the query needs to be …

indexedExpiryDate :[NOW/DAY-5DAY-30MINUTES to *]  … to get all entities with indexedExpiryDate  >= today

Useful links …

Solr Date Field

Solr Date Query Syntax

Time Zone Converter

UTC – Coordinated Universal Time

Say you have

Order class with an property which is Product class type or List of OrderItem objects … instanceof match will fail (even if you have your own Custom Converter) … as the instance will be OrderItems_javaassist_xxx etc. (even if you access the properties to load the properties).

The simple solution is to explicity invoke .toXML for the property. I found more complex solutions using reflection here. I went with the first approach.

If you find better solutions do let me know.

Hibernate was very a tempting option given we already use Hibernate as the ORM. It just seems so elegant and simple. Also Hibenate takes care of updation to the index and not requiring us to explicitly run that … making the index near real time.

But in the end we went with SolrJ … because it supports Faceted Search results and Geo Spatial search results. (Later can be used for LBS and is a feature of Solr 3.1).

(Hibernate Search supports 1:1 facets and not 1:M facets … )

I guess SolrJ will always stay more current and cutting edge than Hibernate Search.

Link to SolrJ

Link to Solr

Link to Solr Geo Spacial Search

String queryString = “select product.productName as productName, count(billQuantity) as countBillQuantity, sum(billAmount) as totalBillAmount from bill group by product.productName”;

Query queryObject = getSession().createQuery(queryString);

queryObject.setResultTransformer(Transformers.aliasToBean(QueryResultSummary.class));

QueryResultSummary is a class with attributes listed below and appropriate getters/setters

- String productName

- Long countBillQuantity (Count always returns Long)

- Float totalBillAmount

Rest is the same as any Hibernate query …

queryObject.list();

Now you can access them from the JSP (or as Struts properties) like with persistent classes.

Keep in mind the Transformer behaves funny with null values.

It’s just a dumb mistake of missing writer.endNode  not matching writer.startNode

Wouldn’t this be common error that deserved better messaging ??