Really odd module behaviour

This is very strange.
First, this works ok:
dbxml> setNamespace x http://www.example.com
Binding x -> http://www.example.com
dbxml> query 'declare function x:foo() as element() {<foo><bar></bar></foo>}; x:foo()'
1 objects returned for eager expression 'declare function x:foo() as element() {<foo><bar></bar></foo>}; x:foo()'
dbxml> print
<foo><bar/></foo>
Now lets do the same thing in a module:
doug@andari:tmp$ cat > x.xquery
module namespace x = "http://www.example.com";
declare function x:foo() as element() { <foo><bar></bar></foo>};
doug@andari:tmp$ dbxml
dbxml> setBaseUri "file:///home/doug/tmp/"
Base URI = file:///home/doug/tmp/
dbxml> query 'import module namespace x = "http://www.example.com" at "x.xquery"; x:foo()'
1 objects returned for eager expression 'import module namespace x = "http://www.example.com" at "x.xquery"; x:foo()'
dbxml> print
<foo/>
That's definitely wrong. I'm getting the same behaviour with all sorts of modules that worked perfectly well on 2.3.10. Since the forums aren't full of complaints, I'm guessing this is only affecting me. I've built the software on two machines though, and I get the same behaviour.
Any ideas how to track down what's causing this?
Cheers,
Doug.

Hi John,
I applied the patch and it fixed the issue. Thanks.
But I have encountered another issue related to import module. I have written up a simple test case to illustrate the problem.
I created two files, test01.xqm and test02.xqm, and put them in the same directory.
======= test01.xqm =========
module namespace test01="http://www.mydomain.com/xquery/modules/test01";
import module namespace test02="http://www.mydomain.com/xquery/modules/test02" at "test02.xqm";
declare function test01:main() as xs:string?
$test02:myString
declare function test01:main2() as xs:string?
test02:main()
========= test02.xqm ========
module namespace test02="http://www.mydomain.com/xquery/modules/test02";
declare variable $test02:myString as xs:string := "test02 has a string";
declare function test02:main() as xs:string?
$test02:myString
==========================
dbxml> setBaseUri "file:///home/lap/bdbxml/debug/"
Base URI = file:///home/lap/bdbxml/debug/
dbxml> query ' import module namespace test01="http://www.mydomain.com/xquery/modules/test01" at "test01.xqm"; test01:main() '
stdin:2: query failed, Error: A variable called {http://www.mydomain.com/xquery/modules/test02}myString does not exist [err:XPST0008], file:///home/lap/bdbxml/debug/test01.xqm:7:3
dbxml> query ' import module namespace test01="http://www.mydomain.com/xquery/modules/test01" at "test01.xqm"; test01:main2() '
stdin:3: query failed, Error: A variable called {http://www.mydomain.com/xquery/modules/test02}myString does not exist [err:XPST0008], file:///home/lap/bdbxml/debug/test02.xqm:7:3
dbxml> query ' import module namespace test02="http://www.mydomain.com/xquery/modules/test02" at "test02.xqm"; test02:main() '
1 objects returned for eager expression ' import module namespace test02="http://www.mydomain.com/xquery/modules/test02" at "test02.xqm"; test02:main() '
dbxml> print
test02 has a string
=================================
Looks like from test01.xqm, it can see test02:main() but not $test02:myString.
Any advice would help. Thanks!
--Lapinski                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           

Similar Messages

  • Really odd First Launch issues

    Hello All,
    I am having som really, really odd issues with my PowerMac G4...
    Everytime I start-up from a cold boot (I turn it off during the night...) I return in the morning, and it can't see my network!
    I usually just restart again, and everything is fine.... but last friday, I had to fight with it for close to an hour to get it back working.
    No patches have been applied, no setting have been changed, it's on a APC UPS system.. it is just used as a workstation...(graphical work)
    It is running 10.3.9... (and has been since it was available, this issue has been going on for about a month now)
    Any clues to this slightly aggrivating problem?
    I don't really want to get a new Desktop until a Commercial MacIntel machine is available.
    Thanks!!!

    Hi Gabriel!
    Why turn it off? Leave it on all night since OS X runs maintenence on the computer during the night. This could be part of problem . I would also suggest booting up from the install disk and running disk utility from it and repairing permissions.
    Also, is it possible your APC UPS system is getting turned off at night too? Perhaps the battery is going bad and just needs replacing?
    Good Luck!
    G4 Dual 450Mhz; 512mb Ram; 40 & 13 gb HD's;Epson 1640; Pioneer DVR-110 DVD/RW   Mac OS X (10.4.4)   BeigeG3/300MT(B) Mac OS 9.2.2; & a Compaq S4300NX;XPsp2;1.5g Ram

  • Really Odd Crash

    There's a really odd bug / crash that happens under these very specific circumstances.  Unfortunately I can't upload an FLA here, so I've provided a download link from a 3rd party site.  If you don't feel comfortable with that, I've listed the steps to recreate the error below.
    (Note: Don't click the giant download buttons, click where it says "Save file to your PC: click here")
    http://www.2shared.com/file/qJWkGney/PrivateVideoChat.html
    http://www.2shared.com/file/M5tNwwRL/PrivateVideoChatExample.html
    If you don't feel comfortable with that, here's a description on how to recreate the error:
    1. Create a new AS3 FLA
    2. Under AS3 Settings -> Library Path, import the flash only player 10 AFCSFlash.swc
    3. Import rpc.swc from the FLEX sdk into the library path
    4. Create a movie clip and put text object in it.  Don't put it on the stage -- leave it in the library.
    And then to make it work with the document class...
    5. Add a Button component and name the instance connectButton
    6. Add a TextInput component and name the instance connectSessionID
    7. Add another TextInput component and name the instance myCSessionID
    8. Create the document class and link it to the FLA
    9. Add your developer credentials (where the constants are defined)
    The document class is this:
    package {
        import com.adobe.rtc.authentication.AdobeHSAuthenticator;
        import com.adobe.rtc.collaboration.AudioPublisher;
        import com.adobe.rtc.collaboration.AudioSubscriber;
        import com.adobe.rtc.collaboration.WebcamPublisher;
        import com.adobe.rtc.collaboration.WebcamSubscriber;
        import com.adobe.rtc.events.SessionEvent;
        import com.adobe.rtc.session.ConnectSession;
        import fl.controls.Button;
        import fl.controls.TextInput;
        import flash.display.Sprite;
        import flash.events.MouseEvent;
        public class PrivateVideoChatExample extends Sprite {
            protected const USERNAME:String = "";
            protected const PASSWORD:String = "";
            protected const ROOM_URL:String = "";
            protected var _cSession:ConnectSession = new ConnectSession();
            protected var _webcamPublisher:WebcamPublisher = new WebcamPublisher();
            protected var _localWebcamSubscriber:WebcamSubscriber = new WebcamSubscriber();
            protected var _audioPublisher:AudioPublisher = new AudioPublisher();
            protected var _audioSubscriber:AudioSubscriber;
            public var connectButton:Button;
            public var connectSessionID:TextInput;
            public var myCSessionID:TextInput;
            public function PrivateVideoChatExample() {
                var authenticator = new AdobeHSAuthenticator();
                authenticator.userName = this.USERNAME;
                authenticator.password = this.PASSWORD;
                this._cSession.roomURL = this.ROOM_URL;
                this._cSession.authenticator = authenticator;
                this._cSession.login();
                this._cSession.addEventListener(SessionEvent.SYNCHRONIZATION_CHANGE, this.onLogin);
                this.connectButton.addEventListener(MouseEvent.CLICK, this.connectTo);
            protected function connectTo(event:MouseEvent) {
                if(this._cSession.userManager.getUserDescriptor(this.connectSessionID.text)) {
                    this.createNewPrivateChat(this.connectSessionID.text);
            public function onLogin(event:SessionEvent):void {
                this.myCSessionID.text = this._cSession.userManager.myUserID;
                this._audioPublisher.publish();
                this._webcamPublisher.publish();
                this._localWebcamSubscriber.subscribe();
            public function createNewPrivateChat(userID:String) {
                this._audioSubscriber = new AudioSubscriber();
                trace("Hasn't crashed yet....!");
                this._audioSubscriber.subscribe();
                this._audioSubscriber.publisherIDs = new Array(userID);
                trace("Aaaaannnndddd...boom.");
    You'll notice if you do any of the following it'll work fine:
    1. Replace the player 10 AFCSFlash.swc with the player 9 AFCSFlash.swc in the library path
    2. Remove the rpc.swc from the library path
    3. Remove the MovieClip with the text object in it
    4. Comment out the line "this._audioSubscriber.publisherIDs = new Array(userID);"
    Edit: Oh yeah, and to use the application you copy your session ID (from the myCSessionID TextInput) and paste it into the the connectSessionID TextInput in another instance of the swf file (or into the same one -- you get the same results).
    I was able to replicate this on another computer using the files above, so I don't think it's my environment that's causing the error.
    I have this version of Flash installed: 10,1,53,64

    REast wrote:
    > As a workaround, I have installed Firefox addon
    "Flashblock"
    > (https://addons.mozilla.org/en-US/firefox/addon/433)
    which prevents all flash
    > content from starting unless I want it to. Kind of
    annoying to have to do this,
    > but it means that the debug flash player is less likely
    to crash my browser
    > straight away.
    >
    >
    All,
    This looks like a bug with Flash player debug plugin.
    Please download the Flash player latest version of debug
    plugin 9.0.115
    from the location
    http://www.adobe.com/support/flashplayer/downloads.html
    The link should show - 12/3/2007 - Update Linux Debugger
    versions (aka
    debug players or content debuggers)... Download the linux
    debugger and
    standalone players.
    You dont need standalone player though it comes bundled
    together. After
    you untar the folder you will see /plugin/ directory which
    contains
    libflashplayer.so
    copy this to your $HOME/.mozilla/plugins/ directory... for
    root users it
    is /usr/lib/firefox* ...
    Relaunch the browser and confirm if you are running Debug
    version of FP
    - by visiting this URL
    http://kb.adobe.com/selfservice/viewContent.do?externalId=tn_15507
    Thanks
    Arun

  • 10.6.5 update - 27" iMac, Radeon 5750 odd graphics behaviours?

    I updated my new iMac 27" to OS 10.6.5 today and have seen some odd graphics behaviours.
    1. Dock icon for Activity Monitor becomes transparent. Running AM, hidden showing CPU history
    2. White rectangle at bottom of expanding folder (list view)
    3. OK Button blue/white check mark pattern when configuring VPN. GUI feels "slow"
    I updated using the Software Update tool; noticed these issues and re-updated using the Combo Installer; no change.
    Any ideas or suggestions? Is this just me? I have a MacBook -- no probems at all!

    N Gallagher,
    I had the problem earluer, but it got far worse with 10.6.5, however, I fixed it by updating my wacom tablet driver.
    It wasn't the system update, but an old driver causing the problem.
    Resetting PRAM, SMC and permissions and all the other standard drivel people post is nice sentiments, but it should be clear that the problem is something else because that solution hasn't solved the problem for anyone.
    I have the sneaky suspicion that it's old flash and old drivers for most people.
    See my original post here: http://discussions.apple.com/thread.jspa?messageID=12570261#12570261
    BUT. If you have a Wacom tablet, for to their website and download the latest driver from September 10, 2010.
    That's all it took for all my graphics problems to go away.

  • Odd software behaviour Lumia 720

    I'm the proud owner of a Lumia 720 since July. I'm facing some odd software behaviour recently though. Without notice or any pattern, the current app changes to another one, sometimes four or five times while my phone vibrates like mad. It happens two, or three times a week, and there's no significant pattern but this weird "phone on speed" behaviour always ends in the Bing app. I thought rebooting was the answer but it doesn't. Does someone has any clue?

    Ugura,
    Thanks for your reply. I adjusted the touch sensitivity but that did not do the trick. Last night, I was already sleeping, suddenly my Lumina 720 played a song that I played before many weeks ago. That was quite odd too. With my phone connected to the charger, me asleep and there was that terrible noise from an old U2-song... And I could not stop it, nor could I lower the volume so I had to turn on my computer to find the keys to a soft reset and it was in the middle of the night.
    So I waited till this morning to reset my phone to it's factory-settings. Much to my annoyance- without any external app loaded, the phone went on nuts again. This time, while vibrating taking numerous screen shots... It's like my phone is haunted. A little poltergeist in it. I need some high-tec exorcism I'm afraid.
    Hope someone can help me further!

  • Odd SocketPermission behaviour

    Hi,
    I've been trying to use SocketPermissions to restrict the IP addresses from which a ServerSocket will accept a connection and am getting some wierd behaviour.
    The ServerSocket will reject and close the first connection attempt from a disallowed address. Subsequent connections from the same host (using telnet) are not rejected/accepted or closed - they just 'hang there'.
    The SecurityManager is also really slow at deciding whether a connection can be accepted or rejected.
    Is this usual ? Or is there a way to setup SecurityManager to behave 'nicely' ?
    regards
    Tony Seebregts

    Just spent about an hour playing around with various things and now it seems to have resolved itself. Odd... Marking this topic as answered now.

  • Really strange promotion behaviour (Required dependencies not found)

    Hello,
    We have a really mysterious failure with the promotion failure.
    I created a very simple promotion job, that should promote a single WebI report from platform A to B. B doesn't have any folders yet. When I create the promotion job with our Testuser, I get the following promotion window and when I start the promotion job, everything works fine:
    As you can see the folder of the WebI report is added automatically to the promotion objects.
    When I create exactly the same promotion job from another client PC in our company, but with the same Testuser ( ! ), the promotion window looks like this, the folder is not added to the object list.
    After starting the promotion job, I get the following error:
    Übergabestatus=Übergabe versucht und fehlgeschlagen., Hochstufungsstatus=Failure : com.businessobjects.lcm.utilities.LCMException: com.businessobjects.sdk.biar.BIARException$EngineException: Erforderliche Abhängigkeiten im Zielsystem nicht gefunden: '[M1M6uOUACCVhAHUAagAAXIgFAgrH.HcAAAA]'
    -> translated: Required dependencies not found: '[M1M6uOUACCVhAHUAagAAXIgFAgrH.HcAAAA]'
    It's the missing folder in the second screenshot, which has that CUID on the source system.
    We tried it with the same browsers (IE and Firefox), same browser versions, etc.
    As we are executing exactly the same steps on both client PCs, the same user, the reason for the different behaviour must be a different client setting I guess?
    Any ideas what could be the problem?
    Thanks!

    hi Hans,
    1) Can u  try import Wizard / Upgrad management tool?
    2  )What Bo version and What patch level you are on? Does the Production/ Dev and and your UAT system are on the same patch level?
    Please try these in Dev Enviorment...
    goto :
    "http://myBOESERVERname:8080/AdminTools" using
    One can use the query builder tool ti figure out which object is causing the issue,
    SELECT * FROM CI_SYSTEMOBJECTS WHERE SI_CUID = 'M1M6uOUACCVhAHUAagAAXIgFAgrH.HcAAAA'
    I Really think the isue is regarding access rights...

  • Odd folder behaviour in Mail 2.0.7

    Hi,
    Mail's been behaving oddly lately ... though I've repaired permissions after every install/upgrade etc.. Wonder if anyone had any ideas about this ...
    I have a lot of folders and sub-folders for various clients and other junk as I really don't like an inbox with hundreds of mails to sift through. Much easier for me to separate them out.
    However sometimes folders just disappear temporarily. Quitting mail then re-opening it always brings them back.
    Most often it's 'drafts' or 'outbox' that goes AWOL but sometimes it's one of the folders I've set up. I've just double checked the whole thing and I've not got a lot of data, certainly no where near the 2gb mailbox limit I read about somewhere else.
    Clues? Suggestions? Is it broke?!?
    Thanks ...
    Greg

    Greg,
    With the legacy of prior versions of Mail, and the enhancement applications, there is a good bit of potential for unexpected problems. However, none of those "potentials" come immediately to mind for this problem. However, with Mail 2.0, there is a completely new structure, and approach to indexing everything. The presence of left over files, with is common, can complicate the health of the index, as also might any non-current structure of the Mail folder.
    In the Mail folder, there should be one or more account folders (depending on the number of accounts) and you have mentioned at least two. In the Finder, open Home/Library/Mail and observe. Your NTL POP account and the Gmail account should each have a POP folder. Within each POP folder, will be some files for administration, and also folders named INBOX.mbox, Sent Messages.mbox, Drafts.mbox, Deleted Messages.mbox, and if using Junk filtering in Automatic, a Junk.mbox. Do you find this to be the case. The POP folder should be at the first level of the Mail folder, and subordinate to any other folder hierarchy.
    Also, at the first level of the Mail folder, you will find the Mailboxes folder -- in the Mailboxes folder should be found the Outbox.mbox, and any xxxxx.mbox folders for any mailboxes you have created. These xxxx.mbox folders may be within other folders, and not. Do you find this organization in place? By any chance, do you find any folders named Personal within the Mail folder?
    With regard to leftover files, that can be quite problematic, let me include the following:
    Since your mailboxes, or at least some of them, originated in Panther (and earlier), I want you to read the article at the link below, and get back to me with some observations, before attempting any changes:
    http://docs.info.apple.com/article.html?artnum=301315
    This issue of leftover files applies not only to mailboxes you have previously created (which are in the Mailboxes folder the article mentioned) but can also apply to those mailboxes in account folders created prior to the upgrade to Tiger.
    Also, you may want to report the size of each file or folder found within a typical mailbox such as INBOX.mbox and particularly Drafts.mbox. Do not, however, try to list the files that are within the folders named Messages, that are found in each and every mailbox. If conversion was fairly recent then it may be very important to compare the size of any file named "mbox" with that of the Messages folder. If the mbox is greater in size than the Messages folder, then not all messages have been converted to be in the Messages folder.
    Once again, the path to the Mail folder is Home/Library/Mail -- within the Mail folder are one or more account folders and the Mailboxes folder. In addition, with Mail 2.0, there is a file named Envelope Index at the first level of the Mail folder. This new universal index supplants the separate index file once found in each mailbox -- it health may be impacted by the leftover files, if any.
    More info, please.
    Ernie

  • Odd sync behaviour

    Here's an odd thing - when I connect the Touch iTunes opens and syncs automatically as it is set to do. Some of the playlists of music will recopy to the iPod each time, but others don't. The playlists haven't been changed, but each time they will recopy as if they are new. The only difference between these and the ones that don't is that these have had lyrics added by MPFreaker, whilst the classical music in the lists that don't resync haven't. If I click the box to manage music and videos manually, then each time I want to change even one thing this set of playlists recopy again. Any thoughts?
    Chris

    Answers..
    are Yes Yes and I see the last record process stored in the xmldata object. All looks good from IDM.
    I have located the problem and it is outside IDM control. I think our Sync Policy is correct.
    This database Table is being modified in a very crude manner. It seems that once per minute (faster than our AS poll loop interval) there is a 3rd party system which pushes any 'changes' into the table from a .csv file. THIS process doesnt do any diff or compares first, it simply writes what it has in the file to the table hence changing the timestamps GRRRRRRRRRR lazy b*sterds, and this is causing IDM AS to spin round and round constantly.
    Thankyou for your help. It helped us eliminate IDM as culprit.
    oh and its an MS SQL table and the "timestamp" is a sort of insertion counter, the higher the number the later the data has been created/updated.
    The message really is in the medium.
    GF

  • Odd Quicklook behaviour

    Running mavericks 10.9.2 have noticed a small quirk in the behaviour on one of my installs.
    Selecting a group of images (mostly jpegs, sometimes tiffs) and opening quicklook. scrollling through
    and quite randomly some of the images will momentarily flash up then turn blank, if you cycle through to the start and continue
    sometimes those that didn't show, will show and vice versa. if you select the grid option then all will show.
    all images are in either sRGB or RGB in the most part 8 bit, and all without layers or alpha channels.
    there is another hard drive on my machine that has the same version of mavericks
    and does not display this behaviour
    I have also removed the quick look preferences from the users library and restarted but the behaviour continues.
    so i was wondering if anyone has come across this and if they have any ideas as to why this might be happening
    any known issues with third party apps etc.
    any advice would be gratefully received
    thanks in advance.

    Just spent about an hour playing around with various things and now it seems to have resolved itself. Odd... Marking this topic as answered now.

  • Odd return behaviour

    okay, consider the following example:
    public class Oops {
         public static void main(String args[]){
    1:          String oops;
    2:          return;
    3:          System.out.println(oops);
    }we get compile error at 3 saying unreachable statement .. fair enough.
    now, consider this:
    public class Oops {
         public static void main(String args[]){
    1:          String oops;
    2:          if(true) return;
    3:          System.out.println(oops);
    }we have modified two and now there is no errors whats-so-ever.
    However, the odd behaviour hecomes obvious if we comment out line 2 where we will get "variable oops might not have been initialized." ... so it would seem that it can recognize that if(true) return; will not let the program reach 3 and hence doesn't care if oops is initialized but why don't we get a similiar error as in example one?

    I don't see anything bizarre about this question. What
    are you referring to?Perhaps 'bizarre' was a poor choice of words. It's just that I recall silkm posting a rather similar question before (a question about how the compiler/jvm handles thing rather than what API can one use to make foo.
    No, that's not a possibility. The JLS explains exactly
    why this behavior occurs:
    http://java.sun.com/docs/books/jls/second_edition/html/
    tatements.doc.html#236365
    The reason that the second example compiles is so that
    you can have code that is conditionally compiled such
    as debugging statements. It's not bizarre at all, it's
    entirely pragmatic and practical.Yes, of course, makes sense. (Although the possibility of a compiler optimizing it out is flashed there.)

  • Odd SOAMANAGER behaviour

    Ok, I just managed to get my old Dell D610 replaced, but this new laptop has some odd behaviour with some Solman transactions.  I can get into Solman Web Workcenter no problem, but when I try to open SOAMANAGER, I get one blank IE8 window (looks like it's supposed to be hidden), and then a few seconds later a second normal looking IE8 window opens, with a partial URL.  This appears to be a problem with my local profile - I can log in to someone else's machine and it'll work, and a coworker can log into my laptop and it'll also work.
    This is the URL it directs the second window to.. https://<server>.<domain>.com/sap/public/myssocntl?sap-client=100
    whereas this is where it should be going.  https://<server.domain>.com/sap/bc/webdynpro/sap/appl_soap_management?sap-client=100&sap-language=EN
    Anyone have any ideas here?  If I call my help desk they're going to be clueless on this one..  lol..
    thanks.
    bk

    Maybe you can rename your home folder in c:\documents and settings and login again - thereby recreating your local profile.

  • Odd JMS Behaviour

    With our application deployed to a client, we have been experiencing some odd behaviour with JMS in Oracle 10g (OC4J 9.0.4).
    The application posts multiple non-transactional messages (roughly 400-500Kb per message) to a queue, which an MDB processes correctly everywhere but the client site. At the client site, only the first message in retrieved from the queue by the MDB, as if to suggest the queue was only capable of holding one message and the other messages are not being added or are being dumped. No exceptions are being thrown, which isn't helpnig diagnose the problem.
    The client site has a similar environment to our internal dev/qa environments - Red Hat AS 3.0 server, v1.4.2 jvm, same OC4J and application config.
    Has anyone else experienced this kind of behaviour with JMS?

    Just spent about an hour playing around with various things and now it seems to have resolved itself. Odd... Marking this topic as answered now.

  • Odd gmail behaviour when using mail.app

    Hello there,
    Ever since upgrading to Mavericks, I've been experiencing odd behaviour when sending messages from my gmail account in mail.app. I send a message and up to ten copies of the message appear in mail.app's trash.
    Does anyone know what might be the cause of this?
    many thanks,
    Chidi

    this is a setting in gmail that interfer with mail.app
    the setting is that mail.app is that when you type a mail it will auto save it while you are typing. and gmail will delete the saved onece and the mail.app will create a new all the time
    duno how to turn this off, try by rebuilding the mailbox in mail.app

  • Odd gc behaviour?

    Hi,
    I'm working on a web application (Jetty, running inside JBoss 3.2.1, Redhat Linux, dual Xeon CPUs, hotspot 1.4.2_04 or sometimes 1.4.2_05) whose gc behaviour seems rather odd. I have been watching visualgc while the application is under load and have observed the following:
    Using the default collector:
    * I have turned off explicit GCs
    * Eden is about 400M and is collected whenever it fills up. Under the load test I've been looking at this occurs about every 30 seconds This collection occurs very quickly (normally a few milliseconds). There will normally be around 2M of garbage at the end of that collection which goes straight into the old generation (maybe via s0 and s1, depending on the settings I'm using)
    * The old generation is also about 400M. However, collections of the old generation happen almost every time that a collection Eden occurs, even though the old generation is very far from full. (In the test I've been looking it seems to have a "steady state" of around 100M). Collections of old take a relatively long time, normally around 1 or 2 seconds. This is an unacceptably long time for our application.
    This seems very odd to me as I wouldn't expect the old generation to be collected every time Eden fills up. For our application it would be perfectly acceptable for the old generation to fill up slowly each time Eden is collected, and then for one big GC to occur once that filled up (which would once every hour or two). But re-collecting the same bit of memory in the old generation twice a minute (at a couple of times per minute) is just no good. I don't understand why the old generation is collected whenever Eden is collected.
    I have tried the concurrent old-generation collector and this only made the problem worse. I also tried setting the old-generation threshold for the concurrent collector to 90% and this made no different - the old generation was still collected every time that Eden filled up.
    Strangely enough, when using the parallel (high throughput) collector for Eden, the problem almost went away. I say "almost" because although no collections occured in the old generation occurred until the old generation had filled completely, the very first old generation collection (which sent it back to its "natural" 100M ") seemed to send it back to its old habits: an old generation collection for every Eden collection.
    Does anybody have any idea what might be going on? Could there be something weird about our application? Am I missing some crucial GC option? Have I stumbled across a GC bug? Do I just not understand hwo this stuff all works?
    Any pointers much appreciated, thanks for your time.
    David

    Hi,
    Thanks for responding.
    I started the vm with the following options:
    -Dsun.net.client.defaultConnectTimeout=15000 -Dsun.net.client.defaultReadTimeout=15000 -XX:+DisableExplicitGC -XX:+UsePerfData -Xmn512m -Xms1024m -Xmx1024m -XX:PermSize=64m -XX:MaxPermSize=128m -XX:MaxTenuringThreshold=0 -XX:+PrintHeapAtGC -XX:+PrintGCTimeStamps -XX:+PrintGCDetails -XX:+PrintTenuringDistribution
    (Note that I've set -XX:MaxTenuringThreshold=0 for this test so I could reproduce the problem more quickly. If I set it to something smaller then the problem still occurs, just less frequently).
    I've got loads and loads of output from the GC trace (I ran a load test on it) so I'll just include an excerpt containing some of the full GCs that I am not expecting. Watching the proceedings in visualgc, I noted that most of the GCs I've included below involved a collection of the old generation (for reasons I cannot fathom). However, they are not noted as "Full GC" in the trace. Any help in understanding this much appreciated:
    722.391: [GC  {Heap before GC invocations=72:
    Heap
    def new generation   total 471872K, used 419455K [0x44950000, 0x64950000, 0x64950000)
      eden space 419456K,  99% used [0x44950000, 0x5e2efff8, 0x5e2f0000)
      from space 52416K,   0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
      to   space 52416K,   0% used [0x61620000, 0x61620000, 0x64950000)
    tenured generation   total 524288K, used 103760K [0x64950000, 0x84950000, 0x84950000)
       the space 524288K,  19% used [0x64950000, 0x6aea4338, 0x6aea4400, 0x84950000)
    compacting perm gen  total 65536K, used 26031K [0x84950000, 0x88950000, 0x8c950000)
       the space 65536K,  39% used [0x84950000, 0x862bbdb0, 0x862bbe00, 0x88950000)
    722.392: [DefNew
    Desired survivor size 26836992 bytes, new threshold 0 (max 0)
    : 419455K->0K(471872K), 0.0272900 secs] 523216K->105143K(996160K) Heap after GC invocations=73:
    Heap
    def new generation total 471872K, used 0K [0x44950000, 0x64950000, 0x64950000)
    eden space 419456K, 0% used [0x44950000, 0x44950000, 0x5e2f0000)
    from space 52416K, 0% used [0x61620000, 0x61620000, 0x64950000)
    to space 52416K, 0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation total 524288K, used 105143K [0x64950000, 0x84950000, 0x84950000)
    the space 524288K, 20% used [0x64950000, 0x6affdfc8, 0x6affe000, 0x84950000)
    compacting perm gen total 65536K, used 26031K [0x84950000, 0x88950000, 0x8c950000)
    the space 65536K, 39% used [0x84950000, 0x862bbdb0, 0x862bbe00, 0x88950000)
    } , 0.0280260 secs]
    734.030: [GC  {Heap before GC invocations=73:
    Heap
    def new generation   total 471872K, used 419456K [0x44950000, 0x64950000, 0x64950000)
      eden space 419456K, 100% used [0x44950000, 0x5e2f0000, 0x5e2f0000)
      from space 52416K,   0% used [0x61620000, 0x61620000, 0x64950000)
      to   space 52416K,   0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation   total 524288K, used 105143K [0x64950000, 0x84950000, 0x84950000)
       the space 524288K,  20% used [0x64950000, 0x6affdfc8, 0x6affe000, 0x84950000)
    compacting perm gen  total 65536K, used 26033K [0x84950000, 0x88950000, 0x8c950000)
       the space 65536K,  39% used [0x84950000, 0x862bc538, 0x862bc600, 0x88950000)
    734.031: [DefNew: 419456K->419456K(471872K), 0.0000370 secs]734.031: [Tenured: 105143K->106385K(524288K), 1.2275840 secs] 524599K->106385K(996160K) Heap after GC invocations=74:
    Heap
    def new generation total 471872K, used 0K [0x44950000, 0x64950000, 0x64950000)
    eden space 419456K, 0% used [0x44950000, 0x44950000, 0x5e2f0000)
    from space 52416K, 0% used [0x61620000, 0x61620000, 0x64950000)
    to space 52416K, 0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation total 524288K, used 106385K [0x64950000, 0x84950000, 0x84950000)
    the space 524288K, 20% used [0x64950000, 0x6b134458, 0x6b134600, 0x84950000)
    compacting perm gen total 65536K, used 26033K [0x84950000, 0x88950000, 0x8c950000)
    the space 65536K, 39% used [0x84950000, 0x862bc538, 0x862bc600, 0x88950000)
    } , 1.2283850 secs]
    747.317: [GC  {Heap before GC invocations=74:
    Heap
    def new generation   total 471872K, used 419455K [0x44950000, 0x64950000, 0x64950000)
      eden space 419456K,  99% used [0x44950000, 0x5e2efff0, 0x5e2f0000)
      from space 52416K,   0% used [0x61620000, 0x61620000, 0x64950000)
      to   space 52416K,   0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation   total 524288K, used 106385K [0x64950000, 0x84950000, 0x84950000)
       the space 524288K,  20% used [0x64950000, 0x6b134458, 0x6b134600, 0x84950000)
    compacting perm gen  total 65536K, used 26033K [0x84950000, 0x88950000, 0x8c950000)
       the space 65536K,  39% used [0x84950000, 0x862bc718, 0x862bc800, 0x88950000)
    747.317: [DefNew: 419455K->419455K(471872K), 0.0000380 secs]747.317: [Tenured: 106385K->107602K(524288K), 1.2445540 secs] 525841K->107602K(996160K) Heap after GC invocations=75:
    Heap
    def new generation total 471872K, used 0K [0x44950000, 0x64950000, 0x64950000)
    eden space 419456K, 0% used [0x44950000, 0x44950000, 0x5e2f0000)
    from space 52416K, 0% used [0x61620000, 0x61620000, 0x64950000)
    to space 52416K, 0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation total 524288K, used 107602K [0x64950000, 0x84950000, 0x84950000)
    the space 524288K, 20% used [0x64950000, 0x6b2649f8, 0x6b264a00, 0x84950000)
    compacting perm gen total 65536K, used 26033K [0x84950000, 0x88950000, 0x8c950000)
    the space 65536K, 39% used [0x84950000, 0x862bc718, 0x862bc800, 0x88950000)
    } , 1.2453910 secs]
    761.940: [GC  {Heap before GC invocations=75:
    Heap
    def new generation   total 471872K, used 419456K [0x44950000, 0x64950000, 0x64950000)
      eden space 419456K, 100% used [0x44950000, 0x5e2f0000, 0x5e2f0000)
      from space 52416K,   0% used [0x61620000, 0x61620000, 0x64950000)
      to   space 52416K,   0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation   total 524288K, used 107602K [0x64950000, 0x84950000, 0x84950000)
       the space 524288K,  20% used [0x64950000, 0x6b2649f8, 0x6b264a00, 0x84950000)
    compacting perm gen  total 65536K, used 26033K [0x84950000, 0x88950000, 0x8c950000)
       the space 65536K,  39% used [0x84950000, 0x862bc7a8, 0x862bc800, 0x88950000)
    761.941: [DefNew: 419456K->419456K(471872K), 0.0000370 secs]761.941: [Tenured: 107602K->109018K(524288K), 1.2185450 secs] 527058K->109018K(996160K) Heap after GC invocations=76:
    Heap
    def new generation total 471872K, used 0K [0x44950000, 0x64950000, 0x64950000)
    eden space 419456K, 0% used [0x44950000, 0x44950000, 0x5e2f0000)
    from space 52416K, 0% used [0x61620000, 0x61620000, 0x64950000)
    to space 52416K, 0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation total 524288K, used 109018K [0x64950000, 0x84950000, 0x84950000)
    the space 524288K, 20% used [0x64950000, 0x6b3c6960, 0x6b3c6a00, 0x84950000)
    compacting perm gen total 65536K, used 26033K [0x84950000, 0x88950000, 0x8c950000)
    the space 65536K, 39% used [0x84950000, 0x862bc7a8, 0x862bc800, 0x88950000)
    } , 1.2193330 secs]
    777.527: [GC  {Heap before GC invocations=76:
    Heap
    def new generation   total 471872K, used 419456K [0x44950000, 0x64950000, 0x64950000)
      eden space 419456K, 100% used [0x44950000, 0x5e2f0000, 0x5e2f0000)
      from space 52416K,   0% used [0x61620000, 0x61620000, 0x64950000)
      to   space 52416K,   0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation   total 524288K, used 109018K [0x64950000, 0x84950000, 0x84950000)
       the space 524288K,  20% used [0x64950000, 0x6b3c6960, 0x6b3c6a00, 0x84950000)
    compacting perm gen  total 65536K, used 26034K [0x84950000, 0x88950000, 0x8c950000)
       the space 65536K,  39% used [0x84950000, 0x862bc998, 0x862bca00, 0x88950000)
    777.527: [DefNew: 419456K->419456K(471872K), 0.0000390 secs]777.527: [Tenured: 109018K->93469K(524288K), 1.4284740 secs] 528474K->93469K(996160K) Heap after GC invocations=77:
    Heap
    def new generation total 471872K, used 0K [0x44950000, 0x64950000, 0x64950000)
    eden space 419456K, 0% used [0x44950000, 0x44950000, 0x5e2f0000)
    from space 52416K, 0% used [0x61620000, 0x61620000, 0x64950000)
    to space 52416K, 0% used [0x5e2f0000, 0x5e2f0000, 0x61620000)
    tenured generation total 524288K, used 93469K [0x64950000, 0x84950000, 0x84950000)
    the space 524288K, 17% used [0x64950000, 0x6a497778, 0x6a497800, 0x84950000)
    compacting perm gen total 65536K, used 26032K [0x84950000, 0x88950000, 0x8c950000)
    the space 65536K, 39% used [0x84950000, 0x862bc1a8, 0x862bc200, 0x88950000)
    } , 1.4292680 secs]
    Thanks for your help,
    David

Maybe you are looking for