ResultHandler - OutOfMemory Exception
|
View:
New views
9 Messages
—
Rating Filter:
Alert me
|
 |
ResultHandler - OutOfMemory Exception
by K. Arnold
::
Rate this Message:
I am trying to iterate over a result set of 2million records, for a large bulk load and transformation into a new ODS. It appears that I am getting an OutOfMemoryException because the DefaultResultSetHandler is caching the object in the nestedResultObjects property. Is there some property I should set or statement/ method call I should be using that will allow me to process one line at a time and not have the nestedResultObjects store each object?
My goal: * Grab a row * Send it to be processed * Once processed, move on to the next row. Note: Once a row is processed I no longer need a tie back to the object. I have included the custom ResultHandler, the unit test and the configuration file. Please let me know if you need other information. package com.primetherapeutics.benplanmgr.entity.rxclaim; import org.apache.ibatis.executor.result.ResultContext; import org.apache.ibatis.executor.result.ResultHandler; import org.apache.log4j.Logger; /** * @author kjarnold * */ public class GroupEligibilityResultHandler implements ResultHandler { Logger logger = Logger.getLogger(GroupEligibilityResultHandler.class); int count = 0; public void handleResult(ResultContext context) { if(context.getResultObject() != null) { count++; logger.debug(count); } //context.stop(); } public int getCount() { return count; } } @Test public void getGroupElibibilitiesByResultHandler() { Map<String, String> parameterMap = new HashMap<String, String>(); parameterMap.put("gelThruDate", "1090101"); parameterMap.put("addDate", "1090911"); parameterMap.put("chgDate", "1090911"); parameterMap.put("planDate", "1090101"); try { GroupEligibilityResultHandler handler = new GroupEligibilityResultHandler(); session.select("com.primetherapeutics.benplanmgr.entity.rxclaim.data.GroupEligibilityMapper.getGroupEligibilities", parameterMap, handler); logger.debug(handler.getCount()); } finally { session.close(); } } Here are my mapping files: <?xml version="1.0" encoding="UTF-8" ?> <!DOCTYPE configuration PUBLIC "-//ibatis.apache.org//DTD Config 3.0//EN" "http://ibatis.apache.org/dtd/ibatis-3-config.dtd"> <configuration> <settings> <setting name="multipleResultSetsEnabled" value="false"/> <setting name="defaultExecutorType" value="BATCH"/> </settings> <mappers> <mapper resource="com/primetherapeutics/benplanmgr/entity/rxclaim/data/BenefitMaxSchedule.xml"/> <mapper resource="com/primetherapeutics/benplanmgr/entity/rxclaim/data/GroupEligibility.xml"/> </mappers> </configuration> |
|
|
Re: ResultHandler - OutOfMemory Exception
by Clinton Begin
::
Rate this Message:
The nestedResultObjects is necessary for join mapping. One way to deal with this though, is to use batches of reads as well as writes. Use the pagination facilities and possibly even the proprietary offset/limit features of your database to grab subsets of the results.
Incidentally I'm rewriting the DefaultResultSetHandler to be easier to understand. But I don't see the need for that cache going away anytime soon... Clinton
On Mon, Sep 14, 2009 at 1:47 PM, K. Arnold <akarnold@...> wrote:
|
|
|
Re: ResultHandler - OutOfMemory Exception
by K. Arnold
::
Rate this Message:
Thank you for the reply. Is it possible to make the join mapping cache a configurable attribute on the mapping file? This is my first time using Ibatis and I was confidently informed that we could walk row by row over a result set. I was told to look up the RowHandler interface. It is my understanding that in Ibatis 3 that the ResultHandler has replaced the RowHandler. Was there a specific design decision made not to allow a row by row walk through?
As a work around I was looking into the Plugin feature. It looks like I can Intercept the resultSetsHandler method, and write my own code to process it. In theory I could utilize the DefaultResultSetHandler code and remove the cache feature in my implementation. However I am not sure if I have access to the same constructor parameters at the point of the intercept. There is some concern around the performance of also introducing multiple call backs to the Database for each "set" of data to process. I need to balance the design choice of writing my own resultSetsHandler with the number of sql calls being made to the database via an offset Select call. I appreciate your replies to this subject.
|
|
|
Re: ResultHandler - OutOfMemory Exception
by Clinton Begin
::
Rate this Message:
The difference between RowHandler and ResultHandler is only in name. They did the same thing.
I'd be interested to know if iBATIS 2.x works for you in this regard. As far as I can recall, 2.x had the same design if you invoked the join mapper...
See, as soon as you try to use a JOIN to load a complex object graph, we need to cache the results so that iBATIS knows to append to the parent, rather than instantiate a new object. Clinton
On Tue, Sep 15, 2009 at 9:04 AM, K. Arnold <akarnold@...> wrote:
|
|
|
Re: ResultHandler - OutOfMemory Exception
by K. Arnold
::
Rate this Message:
I have not used iBATIS 2 to do a comparison. It is very well that the assumption given to me about how RowHandler works is false. I am in the process of trying to figure out a good limit size to pick utilizing an AS400 jdbc driver working on a DB2 database.
We have decided that utilizing the select statement with the limit is the better decision choice for us. Now we need to figure out how to optimize it. Any suggestions would be appreciated  Just to make sure I understand your reply for any future work. I don't believe I am using a join mapper. I have a POJO that maps directly to a custom (yet extremely complicated) sql statement. The POJO doesn't have any collections, and is basically stand alone. It is my understanding from your code, that my type of POJO isn't going to utilize the join mapper method, since the next POJO will be unique.
|
|
|
Re: ResultHandler - OutOfMemory Exception
by Clinton Begin
::
Rate this Message:
Ah, then yes, iBATIS 3 might be more aggressive about using the nested objects cache...(even without join mapping), I can't remember if iBATIS 2 would behave the same way. Might be worth a try... if it works in 2, then I'll revisit the design to see if I can get 3 to only cache if it detects a nested result map.
Clinton
On Tue, Sep 15, 2009 at 10:44 AM, K. Arnold <akarnold@...> wrote:
|
|
|
Re: ResultHandler - OutOfMemory Exception
by K. Arnold
::
Rate this Message:
Out of curiosity, is it possible to create my own DefaultSetResultHandler? I wrote an plugin that would intercept the call, and I was going to walk through the result set myself. Unfortunately I'm not sure how to get the parameters that are used in the constructor. Looking at the code it seems to jump through some hoops passing the objects through. Is there a context that contains the references to those that I could use?
|
|
|
Re: ResultHandler - OutOfMemory Exception
by Clinton Begin
::
Rate this Message:
Hmmm... I don't think so. I suppose it would make sense to add that ! :-)
Jira ticket!
On Tue, Sep 15, 2009 at 11:57 AM, K. Arnold <akarnold@...> wrote:
|
|
|
Re: ResultHandler - OutOfMemory Exception
by Clinton Begin
::
Rate this Message:
Okay, even though you didn't have a Jira ticket open for this, as an added bonus of rewriting the entire result handler, there's now an optimized implementation that should be able to handle any number of rows, as long as you're not using nested result mapping (which makes it impossible to avoid).
Clinton
On Tue, Sep 15, 2009 at 4:34 PM, Clinton Begin <clinton.begin@...> wrote: Hmmm... I don't think so. I suppose it would make sense to add that ! :-) |
| Free embeddable forum powered by Nabble | Forum Help |
