Re: Mysql deadlock / timeout - fixed I hope
We found that there is a bug with MySQL 4.0.20 that causes deadlocks in transactions we have upgraded thanks for the responses - Original Message - From: Robert S. Sfeir [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Tuesday, November 30, 2004 5:53 PM Subject: Re: Mysql deadlock / timeout Huh is OJB competing with some other processes trying to update the same table/db? It looks like it's complaining that it can't get a lock on row, and what ever process is not giving it up. Make also sure that all your transactions end by closing the PB instance. I would add some debugging info to see which record is being updated at the time, with a timestamp to tell you when it was happening, and I would try to catch this specific exception and resubmit the update again once or twice before failing, if there indeed is another process updating the DB. Last thought... Is it around a time when you have a DB backup going on? Some backups make the db/table read only, and that could be a problem. HTH R On 11/30/04 12:43 PM, Jin Bal [EMAIL PROTECTED] wrote: Hello I have an app in production that occasionally suffers from what seems to be a deadlock or timout when doing a simple update inside a transaction here is the stacktrace - can anyone shed any light on things?? TIA Caused by: org.apache.ojb.broker.PersistenceBrokerSQLException: java.sql.SQLException: Deadlock found when trying to get lock; Try restarting transaction, message from server: Lock wait timeout exceeded; Try restarting transaction at org.apache.ojb.broker.accesslayer.JdbcAccessImpl.executeUpdateSQL(Unknown Source) at org.apache.ojb.broker.core.MtoNBroker.deleteMtoNImplementor(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.storeAndLinkMtoN(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.storeCollections(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.storeToDb(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.store(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.store(Unknown Source) at org.apache.ojb.broker.core.DelegatingPersistenceBroker.store(Unknown Source) at org.apache.ojb.broker.core.DelegatingPersistenceBroker.store(Unknown Source) at com.buyacar.dtofactory.OJBTransactionImpl.save(OJBTransactionImpl.java:54) at com.buyacar.dtofactory.UserFactory.saveUser(UserFactory.java:28) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Mysql deadlock / timeout
Hello I have an app in production that occasionally suffers from what seems to be a deadlock or timout when doing a simple update inside a transaction here is the stacktrace - can anyone shed any light on things?? TIA Caused by: org.apache.ojb.broker.PersistenceBrokerSQLException: java.sql.SQLException: Deadlock found when trying to get lock; Try restarting transaction, message from server: Lock wait timeout exceeded; Try restarting transaction at org.apache.ojb.broker.accesslayer.JdbcAccessImpl.executeUpdateSQL(Unknown Source) at org.apache.ojb.broker.core.MtoNBroker.deleteMtoNImplementor(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.storeAndLinkMtoN(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.storeCollections(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.storeToDb(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.store(Unknown Source) at org.apache.ojb.broker.core.PersistenceBrokerImpl.store(Unknown Source) at org.apache.ojb.broker.core.DelegatingPersistenceBroker.store(Unknown Source) at org.apache.ojb.broker.core.DelegatingPersistenceBroker.store(Unknown Source) at com.buyacar.dtofactory.OJBTransactionImpl.save(OJBTransactionImpl.java:54) at com.buyacar.dtofactory.UserFactory.saveUser(UserFactory.java:28)
repository.xml performance tuning best practice
Hi I have an OJB project in poroduction which runs fairly well. I am now interested in tuning OJB's performance / efficiency as I know that my vanilla mapping is not as optimised as it should be (i.e. lazy loading etc). Are there any general (I understand that this issue is quite specific to the app concerned) guidelines or best practices on how to get the most out OJB in production TIA Jin
Re: Negative Sequence Numbers?
I'm not convinced i've got it working yet as all i've done is upgraded in
hope of fixing it.
- Original Message -
From: Robert Crawford [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Friday, July 30, 2004 4:10 AM
Subject: Re: Negative Sequence Numbers?
Charles N. Harvey III said:
It means that something isn't set correctly in the database.
This has happened to me many times. It always happens when I turn on
native sequencing (IDENTITY in MSSQL or AUTOINCREMENT in MySql).
Check the PK of your database, it might not be properly set to
increment. Whenever
this isn't set properly I get -1 and -2 for sequence numbers all the
time.
The database is OK -- I tested it directly, and it assigns sequence
numbers correctly.
This is the code; I've added transaction calls for the PB-api, but the
table I'm using this with doesn't have transactions enabled, so those are
effectively just null-ops:
broker =
PersistenceBrokerFactory.createPersistenceBroker(getPbKey());
broker.beginTransaction();
// 1. make sure no user with that username exists
Criteria c = new Criteria();
c.addEqualTo(emailAddress, username);
Query q = QueryFactory.newQuery(getUserInstanceClass(), c);
Collection col = broker.getCollectionByQuery(q);
if (col.size() 0) {
throw new DuplicateUsernameException(username);
}
// 2. create the new user and return its object
User user = (User) getUserInstanceClass().newInstance();
user.setEmailAddress(username);
user.setPassword(password);
user.setFirstName(firstName);
user.setLastName(lastName);
broker.store(user);
broker.commitTransaction();
Here's the class descriptor:
class-descriptor
class=com.kloognome.accesscontrol.ojb.UserInstance
table=ac_users
field-descriptor name=userNbr column=user_nbr jdbc-type=INTEGER
primarykey=true autoincrement=true access=readonly/
field-descriptor name=emailAddress column=email_address
jdbc-type=VARCHAR/
field-descriptor name=password column=password
jdbc-type=VARCHAR/ field-descriptor name=firstName
column=first_name
jdbc-type=VARCHAR/
field-descriptor name=lastName column=last_name
jdbc-type=VARCHAR/
/class-descriptor
I'm using the 1.0.0 release of OJB, if that matters.
I've tried re-creating the table with the auto_increment and primary
key attributes reversed on the table definition; that didn't change
anything. I've tried forcing the userNbr field on the object to zero; that
didn't change anything.
Any ideas?
--
In discussing [terrorists], 'Capturing their hearts and minds' only makes
sense if by that you mean, literally, capturing their hearts and minds and
putting them in mason jars. -- Jonah Goldberg
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Re: Negative Sequence Numbers?
I've been experiencing the same problem but running on just Tomcat. what version of OJB are you using What DB. I was using RC5 and MySQL i've upgragded and have stopped getting the error in my dev env but have yert to test on the integration server. Cheers Jin - Original Message - From: Vesely, Maxim [IT] [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Thursday, July 29, 2004 2:11 PM Subject: RE: Negative Sequence Numbers? I experienced exactly the same problem running in managed environment with JTA transaction. Max. -Original Message- From: Armin Waibel [mailto:[EMAIL PROTECTED] Sent: Thursday, July 29, 2004 8:35 AM To: OJB Users List Subject: Re: Negative Sequence Numbers? Robert Crawford wrote: Armin Waibel said: Robert Crawford wrote: Armin Waibel said: Do you only use the PB-api? This bug is odmg-api related. I only use the PB-api. Then you shouldn't get problems with negative sequence numbers. Can you describe me with some pseudo code an test case to reproduce your problem? Don't you use transaction demarcation or is this code snip from an managed environment with JTA transaction? Armin From memory, the code looks something like this: PersistenceBroker broker = PersistenceBrokerFactory.createPersistenceBroker(pbKey); User user = new User(); user.setEmailAddress(emailAddress); // more setters here, nothing that touches the primary key broker.store(user); At this point the database has a negative number for the primary key and, from what I can recall from my debugging, the user object has zero in the primary key field. Here's the start of the class descriptor: class-descriptor class=com.kloognome.accesscontrol.ojb.UserInstance table=ac_users field-descriptor name=userNbr column=user_nbr jdbc-type=INTEGER primarykey=true autoincrement=true access=readonly/ - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: sequencemanagernative and mysql
At the moment it is not causing side effects as there is no key collision (yet) - I was wondering whether this was expected behaviour. My understanding is that OJB uses a temp negative PK before insert. does it then lookup the db generated one and set it in the persistent object durign the afterStore() operation? Haven't got any code to post right now (will follow shortly) could there be an issue with the cache (i.e not flushing after insert)? Thanks Jin - Original Message - From: Armin Waibel [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Tuesday, July 06, 2004 5:57 PM Subject: Re: sequencemanagernative and mysql Hi Jin, Jin Bal wrote: I'm finding that the PK field with autoincrement are negative numbers - is this normal behaviour or is somehting up? hmm, the negative values are temporary used before the object was written to DB. See here http://db.apache.org/ojb/docu/guides/sequencemanager.html#Identity+based+seq uence+manager Does it cause side-effects in your environment? Could you post some pseudo code to show the problem? regards, Armin TIA Jin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: sequencemanagernative and mysql
Hi Armin I'm not using the ODMG API so I can't confirm this for you, I'm using PB only. The only object that the app CRUD's (via OJB) is the User object. We did notice that this issue is intermittent and stopped for a while but then started up again. - Original Message - From: Armin Waibel [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Wednesday, July 07, 2004 10:53 AM Subject: Re: sequencemanagernative and mysql Hi Jin, Jin Bal wrote: At the moment it is not causing side effects as there is no key collision (yet) - I was wondering whether this was expected behaviour. My understanding is that OJB uses a temp negative PK before insert. does it then lookup the db generated one and set it in the persistent object durign the afterStore() operation? yep! I made some tests with PB-api and ODMG-api and it seems that when using the odmg api + SequenceManagerNativeImpl + main object with single reference, the FK set in the main object be wrong (negative value). Is this your problem too? Haven't got any code to post right now (will follow shortly) could there be an issue with the cache (i.e not flushing after insert)? normally the object will be stored to cache after the afterStore() call. regards, Armin Thanks Jin - Original Message - From: Armin Waibel [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Tuesday, July 06, 2004 5:57 PM Subject: Re: sequencemanagernative and mysql Hi Jin, Jin Bal wrote: I'm finding that the PK field with autoincrement are negative numbers - is this normal behaviour or is somehting up? hmm, the negative values are temporary used before the object was written to DB. See here http://db.apache.org/ojb/docu/guides/sequencemanager.html#Identity+based+seq uence+manager Does it cause side-effects in your environment? Could you post some pseudo code to show the problem? regards, Armin TIA Jin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: sequencemanagernative and mysql
The code that saves the object is simply broker.store(object); thats it. BTW i'm still using RC5. Jin - Original Message - From: Jin Bal [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Wednesday, July 07, 2004 8:25 AM Subject: Re: sequencemanagernative and mysql At the moment it is not causing side effects as there is no key collision (yet) - I was wondering whether this was expected behaviour. My understanding is that OJB uses a temp negative PK before insert. does it then lookup the db generated one and set it in the persistent object durign the afterStore() operation? Haven't got any code to post right now (will follow shortly) could there be an issue with the cache (i.e not flushing after insert)? Thanks Jin - Original Message - From: Armin Waibel [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Tuesday, July 06, 2004 5:57 PM Subject: Re: sequencemanagernative and mysql Hi Jin, Jin Bal wrote: I'm finding that the PK field with autoincrement are negative numbers - is this normal behaviour or is somehting up? hmm, the negative values are temporary used before the object was written to DB. See here http://db.apache.org/ojb/docu/guides/sequencemanager.html#Identity+based+seq uence+manager Does it cause side-effects in your environment? Could you post some pseudo code to show the problem? regards, Armin TIA Jin - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Concurrency question
I think using the optimistic offline lock approach will be sufficient for your needs. There is doc on the OJB site. but basically any update increments a version number which ojb checks and compares to the version it's about to save and throws an exception which indicates that there is a concurreny conflict. You can detect this and tell the user that someone else is editing the same thing and let them decide how to proceed. HTH Jin - Original Message - From: Fiona Magner [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, July 06, 2004 5:00 PM Subject: Concurrency question Hi there: I am just beginning to use OJB in a Web application. The particular web application (when it goes live with OJB) will be required for high availability. I have a question about a concurrency issue that I have come across. Consider the following example. Two users, a and b are accessing the customers table and updating the customers balance. // 1.User A gets customer record 123 (bean) from database. Customer has balance of 0 // 2. User B gets customer record 123 (accountbalance = 0) // 3.User A make changes to bean object using retrieved values eg. c.setAccountbalance(c.getAccountbalance() + 20.00) (customer will now have balance of 20) // 4. User B makes changes to account balance (balance now 20) // 5. User A sends updates to database (stored balance is now 20) // 6. User B sends updates to database (stored balance is now 20 *** should be 40) Note that steps 1, 3 and 5 actually all happen within the one script (as do 2, 4 and 6) but I have just outlined them as separate steps above for clarity. As you can see it is possible for two concurrent users to cause inconsistency in the data. Is it necessary to have a markDirty() step in the script before applying the changes? Also can you tell me if this markDirty() will put a lock on the table until the changes are committed? Like I say this will be a high throughput site and therefore I would like to avoid any lengthy locks on the tables while updates are being carried out. I hope my email is clear, if you require any more info please let me know. Thanks! Fiona - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
sequencemanagernative and mysql
I'm finding that the PK field with autoincrement are negative numbers - is this normal behaviour or is somehting up? TIA Jin
Re: problem with MySQL driver
try dbalias=//localhost:3306/default?reconnect=true - Original Message - From: Ramon Cano Granizo [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED] Sent: Friday, October 24, 2003 8:59 AM Subject: problem with MySQL driver Hello, I'm working with ojb rc4, MySQL 4.0, a jdbc driver from MySQL (mysql-connector-java-3.0.9-stable-bin) and Tomcat 3.3. The problems comes when the application it's sometime without resolving any request (about two hours), then, the next request throw a StackOverflowException, the trace show the driver opening and closing connection, could someone help me? It's something wrong in the repository_database.xml (here is), perhaps jdbc-lever?? jdbc-connection-descriptor jcd-alias=databaseMySQL default-connection=true platform=MySQL jdbc-level=2.0 driver=com.mysql.jdbc.Driver protocol=jdbc subprotocol=mysql dbalias=//localhost:3306/default username=root password= connection-pool maxActive=21 validationQuery= / :-( - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: storing a new object fails
try leaving the PK empty - I thnk that OJB works out whether its a insert or update by the presence of this field There may be a way to force an insert but I can't remember it right now Jin - Original Message - From: Durham David Contr 805 CSPTS/SCBE [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, September 26, 2003 3:49 PM Subject: storing a new object fails I'm writing a routine that clears out a db and reloads a defined test data set. I'm trying to insert new objects but sometimes, not always, OJB will execute an update statement instead of an insert. I'm explicitly setting the id pk's and bypassing the sequencemanager, so I figure that has something to do with it, but I can't explain the discrepancy. Can anyone point me to the likely cause of this? Thanks, Dave - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Partial collection problem
I've had this problem too. The issue is that the result set is being closed prematurely because the eager-release attribute in the connection descriptor was set to true; if you can try setting this to false. I was able to do this without any probs because I'nm using tomcat, the problem appeared because i tried to port to jboss and changed the eager release setting. cheers Jin - Original Message - From: McCaffrey, John G. [EMAIL PROTECTED] To: 'OJB Users List' [EMAIL PROTECTED] Sent: Wednesday, July 09, 2003 3:46 PM Subject: RE: Partial collection problem Rob, I have had this problem to, and posted all the info that I could to try to diagnose the issue, to no avail. I ended up just coding a work around to get what I needed. I noticed that when I ran a query to get the count of expected objects, the count I got back was right, but the collection was always n+1 (just as you described). For the workaround I first get an enumeration of primary keys (because that works as expected), then go through each of the referenced tables, and build a hashmap of the rest of the objects that I need. Then I piece them together at the end, and return a collection of 'Whole' objects. I think its actually faster than the OJB code would have been, because its fewer trips to the DB. I had to do this workaround, even though it is a little clunky, because we are going to production soon, and I wasn't able to solve the original Collection problem. I only have one class with a collection, so I only had to do the workaround once. I am hoping that the issue gets resolved, because I don't want to have to code any more OJB workarounds, and I don't want to abandon OJB when we move to the next project. Are you able to determine what the issue is? Previous postings pointed to JBoss issues, but I have websphere and DB2 (my issue could be my JDBC driver, I haven't tried any others). It seemed to me that a resultSet was being closed early. please post if you resolve the issue -John -Original Message- From: Rob Kischuk [mailto:[EMAIL PROTECTED] Sent: Wednesday, July 09, 2003 7:09 AM To: 'OJB Users List' Subject: RE: Partial collection problem Thanks for the suggestion. I built from the latest code in CVS last night, and still had the same problem. -Rob -Original Message- From: Armin Waibel [mailto:[EMAIL PROTECTED] Sent: Monday, July 07, 2003 7:23 PM To: OJB Users List Subject: Re: Partial collection problem Hi Rob, I think this is an known issue, please try latest CVS. regards, Armin - Original Message - From: Rob Kischuk [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, July 07, 2003 3:31 PM Subject: RE: Partial collection problem This is a big problem - we have a collection of 200 items that we are retrieving, and the first query only retrieves a single item, with each subsequent query retrieving one more item (e.g. - the fifth query result has 5 items). This seems to be either a very nasty error on my part, or a major bug. Can anyone shed some additional light on this? Is it a bug? Is there a workaround? This is a major showstopper for OJB on an application we're working on, and could require some major rework. -Rob -Original Message- From: Rob Kischuk Sent: Thursday, July 03, 2003 9:43 AM To: '[EMAIL PROTECTED]' Subject: Partial collection problem I saw a discussion cross this list a while back regarding this same problem, but with no solution. The problem is this: when I run a query to load a collection of objects that contain other objects, the first call retrieves only the first item in the collection. The second call retrieves a total of 2 items, the third, a total of 3 items, and so on. Obviously, the correct functionality would be to have all of the items in the collection retrieved on the first call. I described the issue in great detail as a part of Defect OJB172: http://scarab.werken.com/issues/curmodule/120/tqk/0/template/ViewIssue.v m/id /OJB172/issuelist/-1/issuelist/1/issuelist/OJB172 http://scarab.werken.com/issues/curmodule/120/tqk/0/template/ViewIssue. vm/i d/OJB172/issuelist/-1/issuelist/1/issuelist/OJB172 But it appears that the defect hasn't been touched since I entered it. Is there a known fix or workaround for it? Thanks! Rob - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Re[2]: RsIterator question
The approach I take is to limit the number of rows returned and page the results. This can be acheived by setting startAt and endAt indexes on the Query object, then you just have to keep track of the page you're on and work out the start and end index depending on how many you'er displaying on each page also means that you don't have to pass iterators etc to the jsp page keeping you're architecture simpler - you just return the 20 long collection of materialised objects HTH Jin - Original Message - From: Alexander Prozor [EMAIL PROTECTED] To: OJB Users List [EMAIL PROTECTED]; Mykola Ostapchuk [EMAIL PROTECTED] Sent: Wednesday, July 02, 2003 3:39 PM Subject: Re[2]: RsIterator question Hello Mykola, Wednesday, July 2, 2003, 5:27:39 PM, you wrote: MO The main reason I'm using getIteratorByQuery is because my DB table is MO pritty big (~100 000 records). Putting all the records into Collection is MO too memory expensive. MO I'll very appreciate any help of dealing with big DB tables. you can use Proxy for your collection. -- Best regards, Alexandermailto:[EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: How Do Others Use OJB With DTO's and Business Objects?
I also wrestled with these design and patterns. In the end I dumped the business object (well I dumped the DTO's actually) because copying between them was very tedious - OJB handles that for you throught the mappings (so why do it again i thought). I don't see anything wrong with this approach as long as you hide OJB from the rest of your layers (after all they can't see your business objects can they?). The benefit of OJB is that it is transparent so your Bo's/DTO'/Vo's (whatever you've called them! ;-) are just pojo's Once i took this approach my development speed increased and I stopped fighting the architecture. I Still however do translations between the view layer and the business delegate layer and use DTO's to transfer the from http requests (after they've been validated and typed cast etc) to the business layer and back again. HTH Jin - Original Message - From: Shane Mingins [EMAIL PROTECTED] To: 'OJB Users List ' [EMAIL PROTECTED] Sent: Tuesday, June 24, 2003 9:06 AM Subject: How Do Others Use OJB With DTO's and Business Objects? Hi All I just wondered how others are implementing their persistence layers using OJB? This is what I am looking at doing at the moment but I am wondering whether or not it makes sense? I have a Data Access Class (DAO) that provides access to a Data Transfer Object (DTO) like: CustomerDAO with methods insert(CustomerDTO), update(CustomerDTO), getByName(String) etc. So the way I am viewing it is, any returned object from the CustomerDAO would be a DTO. And anything passed to it would be a DTO. Within the CustomerDAO, any interaction with OJB (FYI using ODMG) is done via BusinessObjects. So a getByName(String) will query the database using OJB, return a CustomerBusinessObject to the CustomerDAO which will populate a CustomerDTO and return it to caller. An insert with provide a CustomerDTO to the CustomerDAO which will inturn create a CustomerBusinessObject and persist it. An update will provide a CustomerDTO to the CustomerDAO which will inturn get the existing CustomerBusinessObject, update it from the CustomerDTO and persist the changes. Does this sound reasonable? It just seems like an awful lot of copying that becomes quite difficult when an Object contains a collection of other Objects. I have been using the BeanUtils.copyProperties() method to do my copying but when it come to a collection I have to iterate through it otherwise the objects within the collection of the new DTO object [like returned in getByName()] would contain references to BusinessObjects and not DTO's. Any thoughts or comments? Cheers Shane - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Limited collection returned
Thanks for the detailed insight!
OJB172 was the bug I saw on the dev list#
I'll get back to about the eager-release when I get home.
Interestingly I've just realised that this bug began manifesting itself
after I attempted to port my setup to jboss 3.x from tomcat 4.x
I did attempt to back out the changes however I have a feeling that I left
the eager release attr set to true. I'll check later
watch this space
Thanks again
Jin
- Original Message -
From: Gary Richards [EMAIL PROTECTED]
To: OJB Users List [EMAIL PROTECTED]
Sent: Monday, June 02, 2003 3:05 PM
Subject: RE: Limited collection returned
Hi everyone,
Thought I'd offer my current observations on this as I work with Dan and
am running off the same codebase (JBoss 3.2.x, MySql, ODMG etc) and
therefore trying to crack the same problem.
I've been using the eclipse debugger to step through the action of
invoking a query. I have spent most of the recent time observing
private ManageableCollection getCollectionByQuery(Class collectionClass,
Class itemClass, Query query) of the PersistanceBrokerImpl class (line
1293), inparticular the actions of the while loop on line 1335 which reads
as,
while (((endAt == Query.NO_END_AT_INDEX) || (retrievedCount
numberOfObjectsToFetch)) iter.hasNext())
I have witnessed the hasNext() call of this condition to return false
during the second cycle of the loop, even though I now there to be 13 rows
in the database table and 13 elements behind the Iterator. I have determined
this (for those who know eclipse) by editing the detail formatter of the
RSIterator class to read this.size() which shows 13 when I highlight it.
This leads me onto RSIterator.hasNext() (line 216) which I have moved onto
observe as a result.
I noticed that on the second cycle of the iterator, the call from within
hasNext() will issue
hasNext = m_rsAndStmt.m_rs.next(); (line 223)
This throws an expection which is caught by the same method so that false
is returned.
catch (Exception ex)
{
hasNext = false;
}
I have noticed that the message given by this exception ex (which is not
logged, only gets caught) is Operation not allowed after ResultSet closed.
This certainly suggests that something has closed the ResultSet before the
iterator has been fininshed with (exactly what it says in the message ;) ).
This almost always seems to happen after dependant objects have been
retieved as when we set auto-retrieve = false in repository_user.xml for
the appropriate property we get all rows of the single table we query on but
no dependant objects, however when we set auto-retrieve = true we get
only the first row of the single table and all its dependant objects.
Is this possibly linked to the eager-release feature which needs to be
enabled for operation on JBoss, or a side effect of bug OJB172?
Any other thoughts?
regards
Gary
-Original Message-
From: Dan Hanley
Sent: Mon 02/06/2003 14:38
To: OJB Users List
Cc:
Subject: RE: Limited collection returned
Jin
Are you also running in a managed environment (e.g. jBoss)?
I suspect the bug you're referring to is:
Type :Defect
Issue Id :OJB172
Reported by: Rob Kischuk
rkischuk - (rkischuk at gttx.org)
Details:
Platform: PC
Operating system: windows 2000
Summary: Loading composite object leads to truncated collection
Description: Repeated calls to retrieve a collection of objects that each
contain another object are
retrieving truncated results, which are incrementally built to their
proper length using repeated
requests.
Status: New
Priority: Undecided
Severity: Major
Functional area: Setup
... which sounds like it could be the culprit. A colleague of mine is
trying to trace through with the eclipse debugger to see if we can cast any
light on this.
You are not alone, its holding a lot of people up...
Dan
-Original Message-
From: Jin Bal [mailto:[EMAIL PROTECTED]
Sent: Fri 30/05/2003 22:07
To: OJB Users List
Cc:
Subject: Re: Limited collection returned
I'm also having this problem - thank god someone else's said so !
it's been
driving me mad ;-)
i've been using ojb since 0.9.5 and have only noticed it recently.
(sorry
can't be more precise than that)
here's the offending class descriptor
class-descriptor class=com.buyacar.businessobjects.Vehicle
table=vehicle
blah blah properties etc
!-- IMAGES --
collection-descriptor
name=images
auto-retrieve=true
element-class-ref=com.buyacar.businessobjects.VehicleImage
inverse-foreignkey field-id-ref=1/
/collection-descriptor
/class-descriptor
class-descriptor
class=com.buyacar.businessobjects.BACVehicleDetails
table=bac_details
reference-descriptor
Re: Limited collection returned
I'm also having this problem - thank god someone else's said so ! it's been
driving me mad ;-)
i've been using ojb since 0.9.5 and have only noticed it recently. (sorry
can't be more precise than that)
here's the offending class descriptor
class-descriptor class=com.buyacar.businessobjects.Vehicle
table=vehicle
blah blah properties etc
!-- IMAGES --
collection-descriptor
name=images
auto-retrieve=true
element-class-ref=com.buyacar.businessobjects.VehicleImage
inverse-foreignkey field-id-ref=1/
/collection-descriptor
/class-descriptor
class-descriptor class=com.buyacar.businessobjects.BACVehicleDetails
table=bac_details
reference-descriptor
name=vehicle
class-ref=com.buyacar.businessobjects.Vehicle
auto-retrieve=true
foreignkey field-id-ref=2/
/reference-descriptor
/class-descriptor
whenever I load a BACVehicleDetails (and therefore a Vehicle) I only ever
get one image in the collection.
more over I have a search engine that only returns 1 result the first time
then 2 on the second submit then 3 ..etc :-(
heres the search engine code
PersistenceBroker broker = null;
try {
broker = getBroker();
Criteria criteria = new Criteria();
// check if theyve specified a price bracket
if(searchParams.getMaxPrice() !=null
searchParams.getMinPrice()!=null) {
criteria.addBetween(price,searchParams.getMinPrice(),searchParams.getMaxPr
ice());
} else if(searchParams.getMaxPrice()!=null
searchParams.getMinPrice()==null) {
criteria.addLessOrEqualThan(price,searchParams.getMaxPrice());
} else if(searchParams.getMaxPrice()==null
searchParams.getMinPrice()!=null) {
criteria.addGreaterOrEqualThan(price,searchParams.getMinPrice());
}
if(searchParams.getManufacturer() !=null
searchParams.getModel()==null) {
criteria.addEqualTo(vehicle.model.manufacturerId,searchParams.getManufactu
rer());
}
if(searchParams.getModel() !=null) {
criteria.addEqualTo(vehicle.modelId,searchParams.getModel());
}
if(searchParams.getBodyStyle() != null) {
criteria.addEqualTo(vehicle.derivative.bodyStyleId,searchParams.getBodySty
le());
}
if(searchParams.getFuelType() != null) {
criteria.addEqualTo(vehicle.derivative.fuelId,searchParams.getFuelType());
}
if(searchParams.getOrderBy() !=null) {
criteria.addOrderByAscending(searchParams.getOrderBy());
}
if(searchParams.getChannel()!=null
!searchParams.getChannel().equals()) {
criteria.addEqualTo(channel,searchParams.getChannel());
}
criteria.addEqualTo(status,searchParams.getStatus());
criteria.addEqualTo(visibility,searchParams.getVisibility());
criteria.addGreaterOrEqualThan(displayTo,new
java.util.Date());
criteria.addLessOrEqualThan(displayFrom,new java.util.Date());
Query query = new
QueryByCriteria(BACVehicleDetails.class,criteria,true);
Collection c = broker.getCollectionByQuery(query);
logger.debug(collection size: +c.size());
Iterator results = c.iterator();
ArrayList list = new ArrayList();
while(results.hasNext()) {
// perform cast now to check that we've got the correct
object type
BACVehicleDetails vDetails = (BACVehicleDetails)
results.next();
list.add(vDetails);
}
return list;
} finally {
if (broker!=null) {
broker.close();
}
}
sorry for the long code post
it seems that when the cache is filled up it works ok
help this is stopping me going live with the site :-(
BTW I did a search on the dev list and noticed a bug opened on the 23rd MAy
but i can't find it now
cheers
Jin
- Original Message -
From: McCaffrey, John G. [EMAIL PROTECTED]
To: 'OJB Users List' [EMAIL PROTECTED]
Sent: Friday, May 30, 2003 5:52 PM
Subject: RE: Limited collection returned
yes, I also noticed that, and I tried pre-fetching, but that didn't help
me
either.
so are you having the same problem then? You are only getting one row
back,
instead of many?
(my collection is filled properly with the right child elements, but I am
not getting all of the parents that match the query, and the query is
select
all parents)
can you see the SQL? I couldn't get P6Spy to work, so I don't know what
the
sql looks like.
thanks for speaking up, I wonder if anyone else has had/is having this
problem!
-John
-Original Message-
From: Dan Hanley [mailto:[EMAIL PROTECTED]
Sent: Friday, May 30, 2003 11:47 AM
To: OJB Users List
Subject: RE: Limited collection returned
