Curious Behavior regarding timeouts

hi,
In preparation to writing a bpel process which will be using a synchronous non-bpel webservice call, I wrote a dummy asynch bpel process which does nothing but loop a webservice call that waits for 5 or so minutes... causing a timeout. I catch all, and have an empty element which ignores. Each time the while loop( while(true) ) is excecuted, I do a checkpoint(), and print to the console a message to give me a good vibe that the bpel process is still alive.
Ok.. not good to do, but I'm interested in the behavior of this scenario. My local BPEL PM goes for a few hours just like I expect, then comes up with this:
javax.ejb.EJBException: Transaction was rolled back: timed out; etc...
Not unexpected... just don't know why this happened hours after I started my timeout process... and then I'm puzzled why my process seems to stop functioning.
Then I get:
<2007-07-15 00:48:55,623> <INFO> <default.collaxa.cube.engine> <CubeEngine::expi
reActivity> Instance/activity not found, next expiration attempt (1/5) for activ
ity "1702-BpWai0-BpSeq1.157-3" scheduled for "July 15, 2007 12:50:55 AM EDT"
<2007-07-15 00:50:58,320> <INFO> <default.collaxa.cube.engine> <CubeEngine::expi
reActivity> Instance/activity not found, next expiration attempt (2/5) for activ
ity "1702-BpWai0-BpSeq1.157-3" scheduled for "July 15, 2007 12:52:58 AM EDT"
<2007-07-15 00:52:58,603> <INFO> <default.collaxa.cube.engine> <CubeEngine::expi
reActivity> Instance/activity not found, next expiration attempt (3/5) for activ
ity "1702-BpWai0-BpSeq1.157-3" scheduled for "July 15, 2007 12:54:58 AM EDT"
<2007-07-15 00:54:58,916> <INFO> <default.collaxa.cube.engine> <CubeEngine::expi
reActivity> Instance/activity not found, next expiration attempt (4/5) for activ
ity "1702-BpWai0-BpSeq1.157-3" scheduled for "July 15, 2007 12:56:58 AM EDT"
<2007-07-15 00:56:59,359> <INFO> <default.collaxa.cube.engine> <CubeEngine::expi
reActivity> Instance/activity not found, next expiration attempt (5/5) for activ
ity "1702-BpWai0-BpSeq1.157-3" scheduled for "July 15, 2007 12:58:59 AM EDT"
<2007-07-15 00:59:02,576> <INFO> <default.collaxa.cube.engine> <CubeEngine::expi
reActivity> Exceeded maximum number of expiration retries (5) for activity "1702
-BpWai0-BpSeq1.157-3"
And then my process seems to do nothing else, and is just listed as "open.running" in the BPEL Console... but seems to not "move" any more.
I want to make a bullet proof bpel process which may call a non-bpel (sync) webservice that can perhaps take longer than the synch timeout value on our prod bpel servers. If a timeout occurs, I want to loop (and wait for the scene to cool down (in the version I want to write, the only reason it takes longer than 10 or so seconds is that an unknown number of threads would be accessing the same resource, a ContentDB, and it seems that the more accesses, the slower the response ) ) and keep retrying to call this service after a wait. It seems that out of the box, BPEL does not support this intuitively.... I'm using V10.2.0.2 of the PM.
Oh.. I know that you can change the time that creates this container exception. I do plan on bumping up the values on the deployed domain, but I don't want to rely on this workaround. I don't know the max time to set to work in all situations.
I need to be sure that BPEL will continue to process after this "event" occurs. If a catch all doesn't do it, I don't know what will. It seems that this 'transaction rolled back' is uncatchable? And what is rolled back, and where? Why doesn't it start again at the last checkpoint? I'd appreciate any advise.
Thanks!

I also did this for my webhosting provider, not knowing I could keep it to Welcome. I should've thought of coming here, but I was assuming my errors were because it was Index, not index, so I went in with a text editor and changed all instances of Index with find and replace (names of the programs were changed to reflect it as well). Then I came here, so I edited the Index to Welcome in the iWeb interface, then saved it to the same folder after changing the old folder names to Welcome for an overwrite. It put the new stuff in a new folder of the same name in the website folder. Im using classicFTP. I hadn't checked to see that the folder had been placed inside the old folder, and saw it after I started uploading, when it was too late. So, I uploaded again, this time the inner folder only. My website hasn't changed. Sorry, I know just enough about script to be dangerous to my programs. Can anyone give me a gentle nudge or something in the right direction? I also tried using the question mark but it just gave me a page not found error.

Similar Messages

Very odd behavior regarding variable naming with DataService.fill()

I'm experiencing this odd behavior in LCDS 2.5, wich pertains
to variable naming for the objects returned by a DataService.fill()
call. I ceated a simple destination with a Java adapter. the Server
side Java data object is defined as follows)
package test;
public class RepositoryObject
private String m_strObjectId;
private boolean m_bIsValid;
private long m_lSize;
[public setters and getters here]
On Flex side, the ActionScript value object is defined as
follows:
package test
[Bindable]
[RemoteClass(alias="test.RepositoryObject")]
public class RepositoryObject
public var m_strObjectId:String;
public var m_bIsValid:Boolean;
public var m_lSize:Number;
public function RepositoryObject()
The FDS destination definition (in
data-management-config.xml):
<destination id="testDs">
<adapter ref="java-dao" />
<properties>
<source>test.TestDS</source>
<scope>application</scope>
<metadata>
<identity property="m_strObjectId"/>
</metadata>
<network>
<session-timeout>20</session-timeout>
<paging enabled="false" pageSize="10" />
<throttle-inbound policy="ERROR" max-frequency="500"/>
<throttle-outbound policy="REPLACE"
max-frequency="500"/>
</network>
<server>
<fill-method>
<name>getObjects</name>
</fill-method>
</server>
</properties>
</destination>
What I figured while debugging into the client was is the
data returned by the service (I'm using the mx:DataService call)
returns objects whose variable names
do not match my definition:
m_strObjectId becomes objectId
m_lSize becomes size
m_bIsValid becomes isValid
I wonder why that renaming??? Basically "m_str" prefix was
stripped and capital "O" became lower case "o" for my first
variable etc.
For instance, in a datagrid cell renderer I could not
reference the returned value as {data.m_strObjectId}, but if I use
{data.objectId}, it works.
Any ideas why this behavior (which I found confusing and odd)
would be greately appreciated.
Robert

The latter, as "getM_strObjectId " is an ugly name... don't
tell me that causes issue...ok kidding, I got you. The
setter/getter names are important in this context.
Thanks a lot.
Robert

[BO XI 3.0] - Strange behavior regarding language settings

Hello,
I found strange behavior BOE in infoview portal. We have Desktop intelligence documents published in repository and when we open report (first time) in infoview we have right format for date and numbers e.g. dd.mm.yyyy, 123.123,00.
When we close this report and open once again the format is en-US e.g. mm/dd/yyyy 123,123.00 and this is wrong. We log out and log in and open this report but no change.
If we restart services in cmc (deski services) then the first time is report ok and when we open again this report the same error.
SW:
Windows server 2003 sp2 - english (regional settings - slovenian)
BOE premium ( BO XI R 3.0 => the same problem in BO XI 3.1)
CMS database => Oracle 10gR2 (nsl lang SLOVENIAN_SLOVENIA.EE8MSWIN1250)
Language settings in infoview we check => use browser locale and slovenian
Thank you for reply!
In BOE R2 works fine...
Regards,
Gregor

Hi Gregor,
Try the following option.
Find the file i8ln.xml in the location
"<install directory of BO>\Program Files\Business Objects\BusinessObjects Enterprise 12.0\win32_x86\scripts\"
Take a backup of the file and make the follwing changes.
Find the following code in the file
activate_languages>
<state value="0"/>
<State value="0" cs_lang="Bulgarian"/>
<State value="0" orig_lang="" cible_lang="Bulgarian"/>*
<State value="0"/>
</activate_languages>
Replace it with:
<activate_languages>
<state value="0"/>
<State value="1" cs_lang="English (Australian)"/>
<State value="1" orig_lang="" cible_lang="English (Australian)"/>*
<State value="1"/>
</activate_languages>
Save the file.
Restart your Desktop Intelligence servers. It might take a bit of time to pick up the changes.
Note: If you start Deski after this change, and go to "Tools > Options", you will see a disabled language button with "English(....)" written on it.
Regards,
Anisa

IDA ALV - different behavior regarding "drill-down" for grouped attributes after upgrade to ERP 6.0 EHP7 SPS5

Hello Experts,
we have upgraded our Dev System to ERP 6.0 EHP 7 SPS5 (coming from SPS4).
We had implement an example with an IDA ALV. On SPS4 we could make a "drill-down" to the single lines of grouped attributes. Now with SPS5 this is not possible anymore if the grouping is done for more than one attribute.
On the following screenshot it scan be seen that the grouping is done over a Peril attribute and the Currency. I can only make a drill-down via the peril attribute, so that the aggregated values per currency are displayed. ON SPS4 it was possible to make a further drill-down for the currency attribute, so that the different single lines were visible which were part of the grouping/aggregation.
SPS5 behavior:
SPS4 behavior:
I did not found any new settings which steers the behavior. Also I did not found any note which gives a hint that this is a bug.
Does anyone know if the behavior for that was changed and why?
Thx & Best regards,
Florian

Hi Florian,
it is related with the currency field and it seems to be a bug. Please open a ticket to get the problem fixed. The component is BC-WD-CMP-ALV.
Cheers
Jens

Rescue and Recovery 4.2 Curious Behavior

I use R&R 4.2 to backup simultaneously to my HDD and to a USB HDD. When
a backup is completed I'm always told that my USB HDD is not bootable, and
asked if I want to make it so. It doesn't matter how I respond, because I am
always able to boot from it.
If I answer to make it bootable, one file fails to be copied after five tries:
Z:\Preboot\OPERA\OUSR600.DAT could not be copied, error 5=access is denied.
I've used Add or Remove Programs to repair the R&R installation with no benefit.
I'm curious why R&R thinks my USB HDD is not bootable (though it is), and why
this one file always fails to be copied (though it exists in E:\Preboot\OPERA with a
date-stamp of July 24, 2008 - a date on which I didn't make a backup - and the
permissions on it are the same as all other files in that folder).
Bob Stockler
G41 - 2886-5TU - Mobile Pentium 4 532 - 3.06 Gh
2 Gb RAM - 80 Gb Toshiba 5400rpm HDD - Win XP Pro

Discovered some additional information. Since the previous "fix", the space on my laptop's hard drive was slowly filling up... to the point where there was less than 200 MB left... then suddenly there would be 1 GB of space... 150 MB... back and forth.
When I uninstalled/reininstalled/uninstalled Rescue and Recovery (R&R) 4.2... I found afterwards that over 35 GB of space had "freed-up". I think what happened is that R&R 3.1 was constantly creating more "backups" and/or "partial failed backups" that were cluttering up my hard drive. When I uninstalled R&R 4.2 the second time, these backups were deleted.
If there is a program that can "wipe" any lingering remnants of the R&R program so I can do a clean install of R&R 4.2, the information would be greatly appreciated!
Thanks again!
Sincerely, ThinkPadMan

Curious behavior

Having finally got 4.1 all sorted out for my workflow, I notice something odd. When I make a conversion from ProRes to ProRes (eg 1080 to 1280 or 29.97fps to 25fps) I get the two instances but only about 50% CPU usage.
But make an H264 form a ProRes file, the CPU ramps up to 95%.
Don't expect any answers.

You may be interested in this article.
Russ

Curious behavior of the index page

I use iWeb to maintain a site composed of an index page and many other pages. Whenever I make a change to any of the other pages and republish, I have to go back and republish the index page by itself, otherwise it will not load on any browser. This no big deal, as long as this continues to work, but I worry that something worst may happen, like never reload.!
Has someone run into this small issue and perhaps has a suggestion to resolve it? Thanks.

I also did this for my webhosting provider, not knowing I could keep it to Welcome. I should've thought of coming here, but I was assuming my errors were because it was Index, not index, so I went in with a text editor and changed all instances of Index with find and replace (names of the programs were changed to reflect it as well). Then I came here, so I edited the Index to Welcome in the iWeb interface, then saved it to the same folder after changing the old folder names to Welcome for an overwrite. It put the new stuff in a new folder of the same name in the website folder. Im using classicFTP. I hadn't checked to see that the folder had been placed inside the old folder, and saw it after I started uploading, when it was too late. So, I uploaded again, this time the inner folder only. My website hasn't changed. Sorry, I know just enough about script to be dangerous to my programs. Can anyone give me a gentle nudge or something in the right direction? I also tried using the question mark but it just gave me a page not found error.

DW 8.0.2 Update

Curious behavior regarding Quicktime active content.
The update deals with Flash elements just fine, but does not
correctly deal
with Quicktime active content, audio or video.
All of the elements (flash and QT) on the page linked below
will work fine
in Firefox.
IE will prompt for Active X permission because the QT audio
file is not
affected at all by the fix. The fix prevents the QT movie
from loading at
all as you will see.
http://www.buddhapada.com/tests/qt-active-content-fails.htm
=========================================
This one does work, but it's a done by hand with a completely
different
process...not the one invoked by the Update.
The really odd thing though, is that in DW Design View...the
page is totally
blank, including some text that is on the page; text which is
displayed in
the browser.
http://www.buddhapada.com/tests/qt-active-content-works.htm
===========================================
This one does work but it had to be manually editing from an
embedded QT
format which is designed to add a separate audio track via an
mp3.
But DW did not automatically detect it as needed "repairing"
and if I
manually click on the convert button it causes it to fail.
http://www.buddhapada.com/tests/qt-active-content-fails.htm

Opps...last link was wrong.
This one does work but it had to be manually editing from an
embedded QT
format which is designed to add a separate audio track via an
mp3. But DW
did not automatically detect it as needed "repairing" and if
I manually
click on the convert button it causes it to fail.
http://www.buddhapada.com/tests/qt-active-content-manual-edit.htm

How to trouble shoot curious charger behavior

Having a power cord charging issue.
1.67 GHz PB G4 15" about 8 years old.
The origional power cord died long ago with a pop and smell of smoke.
I replaced the cord with a powerbook medic brand power adapter that has worked
with out flaw for 6 years.
The curious behavior is that when plugged into the port the PB will sometimes stop taking a charge.
Giggle the cord at the port and it will some times return. When the PB stops charging green light on adapter stays green.
How do I determine if it is the charger or the port in the computer that is faulty?

The other weak link is the barrel-shaped power jack within the Powerbook:
(this image from microdocusa.com is of a G3 Powerbook connector but is similar to the G4's)
With age and stress, the solder joints that hold the thin "legs" to the logic board can loosen and cause intermittent charging failures. If you are handy with ripping apart a PowerBook and resoldering tiny parts, that's usually the sure cure, but not one for the faint of heart.
One person I know simply inserted tiny plastic shims between his loose connector and the logic board to wedge the connector back into constant contact. Decided a low-tech approach, but it greatly extended the life of his PowerBook.

JCO - Connection reset by peer - after 5 minute timeout

Hello
We're using SUP with the JCO Connector. This has been working flawlessly since project inception. Now, after switching to a new data center (Windows Server 2008 Server Enterprise Edition, SP2 on VMWare) we're observing a mysterious behavior with timeouts of idle connections. I hope someone in this forum knows what this might be.
JCO connections originate from SUP servers. When such connections are idle for more than 5 minutes they get disconnected from SAP -- this is in agreement with SAP timeout parameter of 300 seconds.
On our new systems we noticed that requests sent 5min+ (after timeout) show following error:
Thread-8 [22:08:55:343]: [JNI-LAYER] RFC.nativeExecute() before RfcCallReceive(2,"Z_U_ONEMOBILE_SWO_SEARCH",0000000010CDC850,0000000000000000,0000000010C81530,000000000C79CA88)
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeExecute() after RfcCallReceive(2,"Z_U_ONEMOBILE_SWO_SEARCH",0000000010CDC850,0000000000000000,0000000010C81530,000000000C79CA88) = RFC_SYS_EXCEPTION
Thread-8 [22:08:55:344]: [JAV-LAYER] JCO.Client.execute (Z_U_ONEMOBILE_SWO_SEARCH) threw a NON-ABAP exception: com.sap.mw.jco.JCO$Exception: (104) RFC_ERROR_SYSTEM_FAILURE: CPIC-CALL: 'CMRCV : rc=20
ERROR       connection to partner '---removed!---:3299'
            broken
TIME        Fri Feb 11 18:08:55 201
RELEASE     710
COMPONENT   NI (network interface)
VERSION     39
RC          -6
MODULE      nixxi.cpp
LINE        4448
DETAIL      NiIWrite: P=161.88.23.170:3299; L=0.0.0.0:49730
SYSTEM CALL WSASend
ERRNO       10054
ERRNO TEXT WSAECONNRESET: Connection reset by peer
COUNTER     3
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect()                                       enter, [SUCCESS]
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect() before RfcClose(2)
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect() after RfcClose(2)
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect()                   with rc = RFC_OK   leave, [SUCCESS]
Our suspicion is that the JCO client doesn't notice the fact that it gets disconnected. We used Network Monitor 3.4 to look at network traffic. Indeed, after 5 minutes the client gets a single frame (with reset flag) and no acknowledgement goes back.
What could this be?
We use
JCO 2.1.8 AMD 64-bit (contains librfc 6.40) and
JCO 2.1.9 AMD 64-bit (contains librfc 6.40) but replaced this with librfc 7.10 (after seeing Note 825494). Behavior was exactly the same. We did confirm in jco/rfc trace that those versions were actually used.
Any ideas?
Thanks and regards
Edited by: B. Gottipati on Feb 14, 2011 9:17 PM

Hello
We're using SUP with the JCO Connector. This has been working flawlessly since project inception. Now, after switching to a new data center (Windows Server 2008 Server Enterprise Edition, SP2 on VMWare) we're observing a mysterious behavior with timeouts of idle connections. I hope someone in this forum knows what this might be.
JCO connections originate from SUP servers. When such connections are idle for more than 5 minutes they get disconnected from SAP -- this is in agreement with SAP timeout parameter of 300 seconds.
On our new systems we noticed that requests sent 5min+ (after timeout) show following error:
Thread-8 [22:08:55:343]: [JNI-LAYER] RFC.nativeExecute() before RfcCallReceive(2,"Z_U_ONEMOBILE_SWO_SEARCH",0000000010CDC850,0000000000000000,0000000010C81530,000000000C79CA88)
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeExecute() after RfcCallReceive(2,"Z_U_ONEMOBILE_SWO_SEARCH",0000000010CDC850,0000000000000000,0000000010C81530,000000000C79CA88) = RFC_SYS_EXCEPTION
Thread-8 [22:08:55:344]: [JAV-LAYER] JCO.Client.execute (Z_U_ONEMOBILE_SWO_SEARCH) threw a NON-ABAP exception: com.sap.mw.jco.JCO$Exception: (104) RFC_ERROR_SYSTEM_FAILURE: CPIC-CALL: 'CMRCV : rc=20
ERROR       connection to partner '---removed!---:3299'
            broken
TIME        Fri Feb 11 18:08:55 201
RELEASE     710
COMPONENT   NI (network interface)
VERSION     39
RC          -6
MODULE      nixxi.cpp
LINE        4448
DETAIL      NiIWrite: P=161.88.23.170:3299; L=0.0.0.0:49730
SYSTEM CALL WSASend
ERRNO       10054
ERRNO TEXT WSAECONNRESET: Connection reset by peer
COUNTER     3
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect()                                       enter, [SUCCESS]
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect() before RfcClose(2)
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect() after RfcClose(2)
Thread-8 [22:08:55:344]: [JNI-LAYER] RFC.nativeDisconnect()                   with rc = RFC_OK   leave, [SUCCESS]
Our suspicion is that the JCO client doesn't notice the fact that it gets disconnected. We used Network Monitor 3.4 to look at network traffic. Indeed, after 5 minutes the client gets a single frame (with reset flag) and no acknowledgement goes back.
What could this be?
We use
JCO 2.1.8 AMD 64-bit (contains librfc 6.40) and
JCO 2.1.9 AMD 64-bit (contains librfc 6.40) but replaced this with librfc 7.10 (after seeing Note 825494). Behavior was exactly the same. We did confirm in jco/rfc trace that those versions were actually used.
Any ideas?
Thanks and regards
Edited by: B. Gottipati on Feb 14, 2011 9:17 PM

STMS error - Timeout during allocate / CPIC-CALL: 'ThSAPCMRCV'

Dear Experts,
We are facing issue in releasing or importing TRs.
1.When releasing TRs in DEV server, it is showing error message "Could not start control program TP" on status bar. But repeating the same process went successful.
2. While Importing TRs into QAS, at the time of refreshing queue/adjusting import queue, it pops-up error " timeout during allocate / CPIC-CALL: 'ThSAPCMRCV' .But after repeating same process, TRs were impoted successfully.
We had performed the following in DEV & QAs:
1. Upgraded kernel to patch level 137.
2. After that replaced TP & R3trans with latest patch levels 169.
3. Re-configured STMS
4. Restarted Server.
Also checked,
(i). Report RSTPTEST in se38- sometimes successful; sometimes error :"timeout during allocate / CPIC-CALL: 'ThSAPCMRCV' (ii). RFC "CALLTP_WindowsNT" in sm59 Connection is OK.
Please help.
Thank you in advance.
Regards

Hi,
Check below link if it helps:
Re: STMS
Pls also check below things :
- check that the "Gateway Options" are maintained in SM59 and the sapdp00/sapgw00 are maintained in the etc/services files. SAP Note 52959
- Please check SAP Note 773830 if you are on a cluster environment.
For the error regarding timeout during allocate / CPIC-CALL: 'ThSAPCMRCV' please check SAP Notes 581509, 566566.
Thanks,
Rupali

MacOS how to use standard UNIX behavior for new files

Hi,
I have one problem. then i create new files and folders on my Mac, they get group owner from parant. How can i say to OS that it should get group owner for new files from user GID? same as it works on UNIX/Linux?
How it works on Mac:
MacOS:/ napetrov$ mkdir -p /tmp/new_dir
MacOS:/ napetrov$ id
uid=34148(napetrov) gid=45956(users)
MacOS:/ napetrov$ ls -al /tmp/new_dir
total 0
drwxr-xr-x 2 napetrov wheel 68 Aug 13 19:06 .
drwxrwxrwt 14 root wheel 476 Aug 13 19:06 ..
How it works on linux:
-bash-3.2$ mkdir -p /tmp/new_dir
-bash-3.2$ id
uid=34148(napetrov) gid=45956(users)
-bash-3.2$ ls -al /tmp/new_dir
total 20
drwxr-xr-x 2 napetrov users 4096 Aug 13 19:01 .
drwxrwxrwt 22 root root 12288 Aug 13 19:01 ..
Message was edited by: napetrov

napetrov, there are two varieties of UNIX behavior regarding the group of a newly created file. The first variety, which Linux uses, comes from System V UNIX — where the file is created with the group ID of the creating user. The second variety, which Mac OS X uses, comes from BSD UNIX — where the file is created with the group ID of the directory in which it is located.
In some UNIXes that come with the System V behavior by default, the BSD behavior can be applied to a particular directory by setting the directory‘s “setgid” bit. However, it is rare for a UNIX that comes with the BSD behavior by default to provide a way to allow the System V behavior for a particular user. To my knowledge, Mac OS X does not provide a way to allow the System V behavior for a particular user; thus, I’d recommend the explicit use of chgrp to set the group ID of a newly created file to that of the creating user, if that behavior is what your application needs.

Question regarding the powershell creation of service applications in SharePoint 2010/2013

Hi
I create all my service applications in SharePoint 2010/2013 using Powershell.
Is there a "best practice" when to start the service Instance for a service application? Is it better to start the service instance prior to creating a service application or after the service application has been create?
I'm curious.
best regards
Bjorn

Not that i'm aware of. If it works then go for it.
I would tend to build the service application and only then activate the instances on servers, however that is going to depend heavily on the application in question. The real heavyweights like Search are a special case in any event.

Stateless session beans and idle timeouts (weblogic 10.3.1)

Need clarification about stateless session beans and the idle-timeout-seconds setting.
Situation is this – we have a process that is timing out due to an API call to a very slow Authentication server.
I am trying to resolve this with a code change, but need to further other understand what the server is actually doing.
We have a session bean which calls another which is calling a util class that does all the work and returns the results back up to the initial session bean. I have left out the ejbName in these examples (it’s irrelevant here).
Example:
SessionBean1 // basically called just once a day
@Session(defaultTransaction = Constants.TransactionAttribute.SUPPORTS,
enableCallByReference = Constants.Bool.TRUE,
type = Session.SessionType.STATELESS,
transTimeoutSeconds = "0",
initialBeansInFreePool = "0",
maxBeansInFreePool = "20")
Methods
@RemoteMethod() public boolean getUserList(String adminGroup) {
Map usrList = getUserList(adminGroup);
Private Map getUserList(String adminGroup) {
     return SessionBean2.getUsers(adminGroup);
SessonBean2
@Session(defaultTransaction = Constants.TransactionAttribute.SUPPORTS,
enableCallByReference = Constants.Bool.TRUE,
type = Session.SessionType.STATELESS,
transTimeoutSeconds = "0",
initialBeansInFreePool = "3",
maxBeansInFreePool = "20")
Method
@RemoteMethod() public Map getUsers(String adminGroup) throws RemoteException {
return javaUtilClass.getUsers(adminGroup);
JavaUtilClass
Method
public Map getUsers(String adminGroup) throws RemoteException {
     // This is where the work happens, calling the Authentication server to get a complete
     // list of users for an admin group. When the user list is around 1500 entries, this can
     // take an hour. Did I mention this server is very slow? It’s about this threshold of 1500
     // that causes the timeout.
     return Map of users
First thought, just bump the idle-timeout-seconds setting for the session beans (from the default 600), but that would be a temporary solution until the user list grew larger.
Second thought, refactor the call to the Authentication Server API to get the user list in blocks of data (say 400 at a time) and decreasing the call/response time between the method getUsers and the API call. This would still occur in the JavaUtilClass, so I am unsure this would make a difference. The session beans would still be idle and subject to timeout, correct?
Would setting initialBeansInFreePool to 1 in SessionBean1 make any difference?
Or should I be looking at replicating the re-factored method from the JavaUtilClass in SessionBean1 where the user list is being used so that the API calls come back to it and keep it 'active'?
Thanks for any advice you could give me on this.

Hi
regarding timeouts, there are two ways:
1.- Changing the settings in the JTA WebLogic domain , called "Timeout Seconds". This will affect globally to all EJB deployed in the domain.
or
2.- Specified directly in the bean with a weblogic annotation, like this:
@TransactionTimeoutSeconds(value = 300)I hope this will help you.
Regards.
Felipe

Timeout Object Problem: Can't recreate scaling error

In my movie there is a function that scales 3D objects over
time using a timeout object. The 3D objects are in an array that is
shuffled each round, and based on the outcome of the array shuffle,
some 3D objects are scaled and some are not.
I have been testing this for a while and it almost always
works as I anticipate, however I have encountered some events
wherein unexpected 3D objects are scaled. To track down the error,
I have been tracing the results of the array shuffle, as this is
what drives the scaling. When I encountered an event where
unexpected objects were scaled I recorded what I thought was a
“bad” array shuffle. Then, disabling the array shuffle
function, I set the shuffle result to my “bad” array. I
expected it would recreate the problem (unexpectedly scaled
objects) however it worked accurately this time around.
Are there any rules regarding timeout objects that are
commonly broken that may be causing this? Does a timeout object
behave any differently in the Director environment than it does in
a published projector? Any broad or general tips (sorry I
can’t post the code) are appreciated.

Hi
In the "fShrink", function the variable called "sTimeout" is
a string:
sTimeout = String("ShrinkTimeOut"+(_system.milliseconds));
in the function "fShrinkLoop", the second parameter being
passed in is the timeout object. Therefore in this function, the
variable called "sTimeout" is a timeout object.
The thing to remember is the object that sends the message
determines what the parameter is (it doesn't matter what you name
the parameter).
In the case of Timeout objects, the first parameter is the
object that Director sends the timeout callback message to (in this
case, your model object... and since your model doesn't have a
method for handling the response, Direct tries the global name
space instead - but still conveniently sends the reference to the
'callback object' as the first parameter). The second parameter is
always a reference to the timeout object itself.
(on a side note - you have taken advantage of a quirk with
timeout objects that let you pass parameters with timeout callbacks
by using the 'callback target' but letting a movie script handle
the callback). If you want to send more than one parameter, you
have to bundle them together into a single paremeter object).
-- Luke

Curious Behavior regarding timeouts

Similar Messages

Maybe you are looking for