Querry about Launchd behaviour

I'm trying to identify the causes of a certain malfunctioning.
I would appreciate advice on this specific issue:
I see this string repeated every ten seconds in Console:
"- com.apple.launchd.peruser.503[121] (com.babylon.activation[775]): posix_spawn("/Library/Application Support/Babylon/Babylon Activation.app/Contents/MacOS/Babylon Activation", ...): No such file or directory"
Of course, I don' thave a Babylon in that place. The question is why launchd calls this application every 10 sec and how to stop it.
I am on a MacBook Pro (8,2), 2011, running MacOSX 10.6.8.
Thanks in advance

It sounds as if you may have a Babylon .plist file somewhere that is responsible for launchd's behaviour, and that .plist is looking for a file in that particular location, which isn't there.
This tutorial may help: http://launchd.info/
A quicker way to track down the culprit would be to do a Spotlight search for Babylon and see what comes up.

Similar Messages

Questions about launchd

I am wanting to schedule backup and since cron is deprecated, I used launchd. I got it to work but I have some questions.
Can I schedule the task without having to reboot my computer?
Can the script go in LaunchAgents or LaunchDaemons? Also, I have seen two different LaunchAgents and LaunchDaemons folders. One for root (I think) and one for user. Which one should I use?
Once I have a task scheduled and want to change the execute times in the plist, do I need to do anything once I make changes to the plist file?
Thanks in advance for the help.
BTW: I am running OSX 10.8

amroge wrote:
Can I schedule the task without having to reboot my computer?
Yes. Use the launchctl command-line tool. Typical usage is "launchctl load -w ..." or "launchctl unload -w ...".
Can the script go in LaunchAgents or LaunchDaemons? Also, I have seen two different LaunchAgents and LaunchDaemons folders. One for root (I think) and one for user. Which one should I use?
Actually, there are three. If it needs to run while a user is logged in, it should go into LaunchAgents. Otherwise, it should go into LaunchDaemons. Use /Libarary/LaunchDaemons. The only issue here is that /Library/LaunchDaemons is for sytem-wide tasks that aren't from Apple. Any scripts you call need to be moved out of any user home directory.
Once I have a task scheduled and want to change the execute times in the plist, do I need to do anything once I make changes to the plist file?
Use launchctl to unload and reload it.
You said you were doing this for backup? Why not use Time Machine?

Question about different behaviour from Jinitator and JRE

Hi,
In the formsweb.cfg, if I set to to use JRE, it works fine.
However, if I change to use Jinitator, it shows the following error in the browser.
error:java.lang.VerifyError :(class:Try1/TryHint$1, method: <init> signature :(LTry1/TryHint;)V) Expecting to......
The sample it works
baseHTMLjinitiator=webutiljpi.htm
baseHTMLjpi=webutiljpi.htm
After setting the below, it shows the above error.
baseHTMLjinitiator=webutiljini.htm
baseHTMLjpi=webutiljini.htm
Anyone can help ?
Thanks.
Ivan

Every config uses the Java Runtime Environment (JRE) -- the difference is which JRE. JInitiator ships with a v1.3 JRE, and the Java Plug-In (JPI) uses whichever JRE corresponds to the given JPI version.
You have given no information regarding which browser/version you are running, and it appears that you have not pasted the entire stacktrace, so it will be very difficult for anyone to help you. Reposting provides no additional information.
Please review the following information -- it will hopefully help you sort things out:
http://fdegrelle.over-blog.com/article-4135746.html
Eric Adamson
Lansing, Michigan

Querry about $PROFILES$

In Aging 4 - Bucket There is almost 23 value sets one of them is XLA_MO_REPORTING_LEVEL and it's value validation is of Table Type.
And the respective code.
Where/ Order By WHERE
lookup_type = 'XLA_MO_REPORTING_LEVEL'
AND ((TO_NUMBER(lookup_code) >=
TO_NUMBER(:$PROFILES$.XLA_MO_TOP_REPORTING_LEVEL)
AND NVL(pg.multi_org_flag,'N')='Y')
OR (lookup_code = '1000'
AND NVL(pg.multi_org_flag,'N')='N'))
ORDER BY TO_NUMBER(lookup_code)
Can any one let me know what is the value retreved by (:$PROFILES$.XLA_MO_TOP_REPORTING_LEVEL)
It'll be help full if any one can provide Simple Example with elaboration.

Check it in
Help->Diagnostics->Examine
Enter block prompt as $PROFILES
Enter field prompth as XLA_MO_TOP_REPORTING_LEVEL
and get its value.
Thanks, Avaneesh

Querry about PCR

could any one send all workflow related documents for process change request in my mail id [email protected] ..as i am new to workflow ...i have to customize the standard workflow for process change request ....
Regards
Vijay

hi vijay,
http://help.sap.com/saphelp_erp2005vp/helpdata/en/c8/a6d002385c4e73baa48aa324f5eedc/frameset.htm
Regards,
Balaji E.

Querry on 1z0-007 exam objectives

I am a student and I have a querry about objective covered in SQL9i (1Z0 -007) Exam. As i had studies course material provided by Oracle Education details are:
(40049GC10 /Production 1.0 / June 2001 / D33052 ) is printed on book.
The detais provided on web site (Test Contetnt Checkkist) is covered till chap 14 and chap 15-20 has objective below
Using SET Operator
1. describe SET operator
2. Use a SET operator to combine multiple queries into a single query
3. Control the order of rows returned.
Oracle 9i Datetime Functions
1. After completing this lession, you should be able use the following datetime functions:
current_date,current_timestamp,localtimestamp,dbtimezone,sessiontimezone,extract,from_tz,to_timestamp,to_timestamp_tz,to_yminterval,tz_offset
Enhancement to the GROUP BY Clause
1. Use the rollup operation to produce subtotal values
2. use the cube operation to produce cross-tabulation values
3. use the grouping function to identify the row values created by rollup and cube
4. use grouping sets to produce a single result set.
Advanced Subquery
write a multiple column subquery
describe and explane the behaviour of subqueries when null values are retrived
write a subquerry in from clause
use scaler subqueries in sql
describe the type of problem that can solved with correlated subqueries
write corelated subqueries
update and delete rown using corelated subqueries
use exist and not exist operators
use the with clause
hierarchical retrieval
interpret the concept of a hierarchical query
create a tree structured report
format hierarchical data
exclude branches from the tree structure
Oracle 9i Extension to DML and DDL Statements
describe the feature of multiple inserts
use the following types of multiple inserts
Unconditional inserts
pivoting inserts
conditional all insert
conditional first insert
create and use external tables
name the index at the time of creationg a primary key constraint
so my main query , is following objectives are included in the list of objectives of 1z0-007 exam or not.
kindly reply to me with proper solution or forward this mail to concern personnel if u r not know abot my query.
thanks and regards,
vikram anand
lucknow (U.P.)
India
[email protected]

There is no such specific Oracle Documentation given in Oracle, rather you will have to refer a couple of related Oracle Documentations.
Here are the Oracle SQL Documentations.
http://download.oracle.com/docs/cd/B10501_01/nav/docindex.htm?remark=homepage#index-SQL
If you need specific document on this exam, you will have to undergo the related training course, then you will be given Oracle Student Guides/ILT Materials.
http://education.oracle.com/pls/web_prod-plq-dad/db_pages.getpage?page_id=41&p_org_id=68&lang=US&p_exam_id=1Z0_007#1
Best of luck.
Regards,
Sabdar Syed.

Launchd: How to Run a Job Only When Waking?

I want to run a script when the computer wakes from sleep. I've read everything I can find about launchd and cannot find a solution.
I tried RunAtLoad, but it did not run the script when the computer woke from sleep.
Is there a parameter for launchd that will run a job only when the computer wakes from sleep?
I know other parameters like StartInterval will run any "queued" jobs when the computer wakes up, but I'm looking for a parameter that runs the job only when the computer wakes up.
Any help is appreciated.
Thanks
Brett

Sorry, but there is no capacity for that built into launchd. You might be able to hack something together (I've tried before, with some limited success), but it will be easier and more stable to use sleepwatcher.

Behaviour of Hidden Key figures in MDX universe

Hello Experts,
I would like to know about the behaviour of Hidden objects in Bex query in an MDX universe. SAP BW environment: BW 7.3 being tried against BI 4.0SP2 and BI 4.0,SP4 and higher.
In a 4.0,SP2 environment, when the display of an object is set to "Always Hide", the universe is still able to recognize the object and show it in measures at universe level.
But the same query when accessed from BI 4.0,SP4 onwards environment, the universe is not able to find the objects that are set to always hide and can see only those objects which are set to either "Always Show" or "Hide(Can be shown").Below is the property i am talking about.
Is this a change in property in BI 4.0,SP04 and higher that if the object needs to be accessed at universe level, it should be set to Always show or Hide(can be shown)?
Thanks and regards,
Abhishek

This is a behaviour of BW 7.3 with BOBJ all versions. Key figures which are Set to Hide(always) cannot be seen in BOBJ , which was the opposite case in BW 7.1 where MDX universes could also recognize always hidden key figures.
Resolution: Change key figure setting to "Hide(Can be shown) to be displayed and recognized in BOBJ all versions which are connected to BW 7.3 and higher.
Closing this thread.

About Check box Prompt in OBIEE 11g

Hi All,
I really wonder about strange behaviour of check box prompt in OBIEE11g.
I have created a check box prompt and associate it with my report. Now when i tried to view result with select any of check box, report run properly, But when i deselect check box and re run report, the check box automatically select again and result of report filtered.
Any one have any Idea about this, as this seems a BUG in OBIEE11g...
Many Thanks.

Hi Archie,
Are you saying, when you deselect a prompt and try to run the report, it still selects the prompts from the first selection and filters the same again?
-Amith.

About switching (newbie question #1a)

Coming from Aperture but trying LR4.1, I've found it painful to learn all the differences when you know Aperture well.
I do have Victoria Bampton's book, which I am slowly going through.
But some initial questions:
-I've seen some definite buggy behaviour (forget what the issue was now), but it went away when I relaunched the s/w:
how stable is LR generally? (being at the 4.1 release level, I thought it should be pretty stable; sounds like the Adobe s/w stability is not near Aperture from the reports I've seen about buggy behaviour (PS on the other hand I've never had a problem with; just upgraded to CS6 which is in part why I also got LR 4.1)
-I don't know if this is a bug or a user issue: in grid view for a folder I select a photo (a virtual version, if that matters), then switch to loupe view:
why does the arrow key (L or R arrow) jump wildly to some distant photo [relative to grid view and relative to filmstrip at botto] and back, instead of just moving along to the next/adjacent photo beside the current photo?
Just tried this again and now it is jumping to the adjacent photo AND BACK: eg keep pressing R arrow and it just toggles between these two adjacent images:
now I must point out that these two photos have a black and white diamond.
But I have loupe view not Compare view. So what is going on?
I just noticed if I clicked on one of the diamonds in the filmstrip I seemed to have exited that funky mode. But then somehow got right back in it. Then noticed I had my second display turned on and turning that off also got me out of the funny mode. This is really confusing.
It appears the secondary display can have its own display mode: eg I've selected Loupe view, but once the secondary display was enabled, it showed Compare view despite my having already chosen loupe view, seems turning on that second display lets LR ignore you're set view...and you have to manually change that second display view also to Loupe to fix the issue.
I think the issue is that the diamonds where on effectively putting me in compare view against my will/my command and against the displayed view mode shown in the View menu (all because the 2nd display was enabled and decided to use its own view mode!!). Does that sound right? Is that a bug or something Adobe would call an undocumented feature or what???
-brushes:
is there a way after I've applied one brush, then hit NEW to apply a second brush, for me to go back and further modify the first brush? EG add to or modify the strength of the first brush effect?
If so, I am not seeing where in the UI you would go to access that.
-keywords:
I imported a keyword list and this shows up fine.
I had a photo and dragged several keywords from this keyword list onto the photo and then let go (when I dragged onto the photo a green circle with a plus sign showed, suggesting I was adding the keyword to the photo).
But the published photo on smugmug showed no keywords.
So I go back to LR and look for the keywords but don't see anything.
Where do an images assigned keywords show up? Is it in the box below I've highlighted in red? I tried again, and after dragging and dropping a keyword on an image, that box remains empty!
Any help appreciated.
-Colin

I'll answer the questions that I can
how stable is LR generally?
Extremely stable for me and others, unstable for some
-I don't know if this is a bug or a user issue: in grid view for a folder I select a photo (a virtual version, if that matters), then switch to loupe view:
why does the arrow key (L or R arrow) jump wildly to some distant photo [relative to grid view and relative to filmstrip at botto] and back, instead of just moving along to the next/adjacent photo beside the current photo?
If you have multiple photos selected, for example one that is visible and one selected that you would have to scroll way down to see in the grid, the arrow keys in loupe mode will behave like this ... it will scroll left or right to the next selected photo, not the next photo
is there a way after I've applied one brush, then hit NEW to apply a second brush, for me to go back and further modify the first brush? EG add to or modify the strength of the first brush effect?
Sure. There is a "pin", actually a circular icon the represents the starting location of each brush. If you click on a pin, you can add additional edits or brush strokes
I had a photo and dragged several keywords from this keyword list onto the photo and then let go (when I dragged onto the photo a green circle with a plus sign showed, suggesting I was adding the keyword to the photo).
But the published photo on smugmug showed no keywords.
So I go back to LR and look for the keywords but don't see anything.
Depends on what you did. Publish to Smugmug, there is probably an option that removes metadata on export. This would be in the Export dialog box.
Go back to LR and the keywords aren't there ... did you by any chance re-import these photos?

Launchd dependencies

I have an service that needs to run 24/7. The problem is that it requires 2 separate processes, the first a server running some java for the backend, and the other an Adobe AIR app drawing images and encoding them for saving. So the server obviously must be up and running first. Then, if either the AIR app or the server crashes, the whole thing should startup again fresh (i.e. if one part goes down, the other app should also get relaunched).
I can get launchd to launch my server with this plist file:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>com.sns.sentinel</string>
<key>KeepAlive</key>
<true/>
<key>ProgramArguments</key>
<array>
<string>/Applications/oc4j/j2ee/home/applications/Sentinel/Sentinel.app</string>
</array>
<key>RunAtLoad</key>
<true/>
<key>UserName</key>
<string>root</string>
</dict>
</plist>
But everywhere I read about launchd it says it doesn't support dependencies and that I "should use IPC". But I have no idea what that means. Can't find an explanation for it anywhere, not even a sample. So how can I ensure that my 2 pronged service stays up and gets relaunched whenever either side goes down?

IPC means Inter-Process Communications. basically, Apple would rather you get your processes talking to each other directly than trying to get them to signal each other through launchd. here's a way-too-unixy link on the topic. in this case, though, I think your best bet would be to write a simple script that checks to see if both processes are running, and relaunches them both if one of them isn't. just have launchd run that script every 30 seconds or so.

Setting up launchd dependencies

I have an service that needs to run 24/7. The problem is that it requires 2 separate processes, the first a server running some java for the backend, and the other an Adobe AIR app drawing images and encoding them for saving. So the server obviously must be up and running first. Then, if either the AIR app or the server crashes, the whole thing should startup again fresh (i.e. if one part goes down, the other app should also get relaunched).
I can get launchd to launch my server with this plist file:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>com.sns.sentinel</string>
<key>KeepAlive</key>
<true/>
<key>ProgramArguments</key>
<array>
<string>/Applications/oc4j/j2ee/home/applications/Sentinel/Sentinel.app</string >
</array>
<key>RunAtLoad</key>
<true/>
<key>UserName</key>
<string>root</string>
</dict>
</plist>
But everywhere I read about launchd it says it doesn't support dependencies and that I "should use IPC". But I have no idea what that means. Can't find an explanation for it anywhere, not even a sample. So how can I ensure that my 2 pronged service stays up and gets relaunched whenever either side goes down?

Start with http://developer.apple.com/macosx/launchd.html and repost to the server products forums or Unix forum (under OS X Technologies).

About the last_call_et column in v$session

Hello,
Can someone help me about a behaviour that is very strange ...
We have connection about SAS on a database which manipulate Oracle table and others SAS Structure.
The query is a "join" about that table and a "SAS Table" between million of lines in the Oracle table one ....
It last a long time ....
During the query, we have time to see at the v$session table...
At any time (it lasts 3h), we saw that the column 'last_call_et' of the v$session never change : it continuesly stay at 0.
What does it mean ? it means that SAS send a query every second ? Or this column shows the last "internal call"?
If this is the right reason, i would'nt understand why, when i take the same query( launched by SAS) in TOAD, there is no the same behaviour and this behaviuor is in that case the right behaviour : the session is ACTIVE but the last_call_et respresents really the time since i launched the query... So why when SAS do it, the last_call_et stay at 0 ? it works like SAS launch every second this query !!!!
Can someone help me ?
Thanks ...

Jonathan Lewis wrote:
user12046632 wrote:
When you say :
A FETCH is a database call :
That means that it must affect the value of 'last_call_et' or not ?"A Fetch is a database call"
"last_call_et" -- certainly sounds like that could be something to do with database calls, perhaps End Time of LAST database CALL.
A parse is also a database call, as is an execute ... I feel that perhaps I was a little too concise with my description. You asked why SAS could be doing lots of calls (although that wasn't exactly how you put the question) and I replied that calls could include fetches. And yes, the last_call_et is reset on every fetch (amongst other things - and ignoring the special behaviour of background processes).Ok
>
(accordint to me, only to user's activity do so and not to "database internal callsl") ! isn't it ?You don't want to believe everything you read on the internet - even if it did come from the online manuals. But a "fetch" is not a "database internal call" anyway, it's a call from a client to the server.Be sure that i don't do so.
But the explanation of Oracle of the term 'cal' is not so precise at all ...
>
Your follonwing explanations seems to answers contrarily ....
But the question is there : what sort of call affetcs the value of "last_call_et"All database calls - which you should read as "all calls from the client to the server".Ok
>
>>
>>
When you say
all_rows plan => parse call, execute call ..... with very long call time, lots of fetch calls
first_rows_n plan => parse call, execute call ... with very short call time, lots of fetch calls
In the first case you would see last_call_et constantly growing until the fetches started
In the second case last_call_et would keep resetting to zero so fast you probably wouldn't see it as a non-zero value.
I don't understand : ok for the conséquences on the timing ... but if there are a lot of fetch call in both cases : the behaviour on the 'last_call_et' has to be equal ?True - but in one case I've included an execute call that is active for a very long time, and in the other I've proposed an execute call that is active for a very short time. (And execute calls are calls, and I did ask you about the exact details of your observations.)Ok
If we consider that every call reset to 0 the value of 'last_call_et',
why, in the case of launching the same request by TOAD, shouldn't i see the reset to 0 when the fetches works (because in TOAD during the whole query, which means since the click on "launching query" to the retrieving of data, the value did not reset to 0) ?
>
>>
When you say
Having said that, I've just run a quick test in SQL*Plus on 11.1.0.6 and 10.2.0.3 to demonstrate this point, and 11g is behaving differently from 10g - in the case where the last_call_et should be growing it stays at zero.
"in the case" : in the case of the v11 or the v10 ?11g
In my case : i'am on a v10.2.0.3 ::: Is there a bug ?Not that I know of.ok. So it is in V11 that you have an incorrect behaviour ....
>>
>
Regards
Jonathan LewisLAST NEWS : we have relaunched the query in SAS and observed that we have the same context of plan in both cases (by SAS en y TOAD) : ALL_ROWS.
Very strange ...
Last question :
Just to be sure : When you talk about "First Rows plan" : it means that the retrieve data are restricted to the first rows encountered
or
it is just the plan of the query which is in "that context" which means that the engine works so although the whole data are retrieved ....
thanks for your return

Irritating behaviour of crop tool

There are three things about the behaviour of of the crop tool in Aperture which I find very irritating compared to, say, the crop (and rectangular selection) tool in GIMP or indeed most other 2D graphics editors:
It's very difficult to crop all the way to the corner of an image. If I drag (e.g.) the lower right corner of the crop rectangle to the lower right corner of the original image, I find that as the mouse cursor moves rightwards past the original right-hand edge, the bottom of the crop rectangle rises(!); similarly, moving the mouse downwards causes the right edge of the crop rectangle to move leftwards (away from the corner)! The only way I can be sure to be cropping all the way to the corner is to separately move the right and bottom edges. (It's as if Aperture is trying to keep the corner of the crop rectangle on the diagonal line between its original position and the current position of the mouse cursor, rather than moving it to the closest position on the boundary of the original image like every other image editor (GIMP, Preview, etc.) does. This behaviour occurs despite "Aspect Ratio: Do Not Constrain". I presume this is supposed to be a feature - but how do I turn it off? It's very, very annoying.)
The handles for the crop rectangle are very small - only 8x8 pixels. In GIMP, the corner handles are something like 32x32 pixels, and the edge handles are similarly wide and as long as the corresponding edge of the image. The larger handles make it much faster to adjust the crop rectangle (due to Fitts's law). Why does Aperture have such unnecessarily hard-to-use controls?
Sometimes, when I go back to an image to adjust a crop which was previously done with "Aspect Ratio: Do Not Constrain", and despite not having changed this option subsequently for this or any other image, I will find that it has become set to some other value (often 5x7). It seems that this happens most often when the the originally-made crop was close to (e.g.) 5x7. Is this supposed to be a feature? If it is, how do I turn it off? I want this option to remain set at "Do Not Constrain" until I explicitly select a different value.
Please: is there any way I can make this essential tool less frustrating to use?

All but the size of the handles are complications which showed up in the most recent upgrade (3.2). There are scores of threads on these. You can (and should) register a report with Apple. From within Aperture, use "Aperture→Provide Aperture Feedback" to get to the problem reporting form.

X-post: Puzzled about RMI

Folks, yes I am cross posting this so try to keep the flames down :) I've also placed it on the RMI forum but no one there seems to want to help and I know everyone in the "Advanced" forum are very helpful (do you like the flattery???).
Anyway, I'm just playing with RMI now and have followed the tutorial but I am a little puzzled about some behaviour I'm observing. This may be that I am missing a piece of information about RMI or how it works so if anyone can fill in the blanks then I would be most appreciative!!!
So I've got the following Remote interface definition:import java.rmi.*;
public interface RMIInterface extends Remote
    public int getValue () throws RemoteException;
    public void setValue (int val) throws RemoteException;
}And I have some code that implements that interface. This code also contains a main method that performs some stuff with the RMI object. It is given below:import java.rmi.*;
import java.io.*;
public class RMIObject implements Remote, RMIInterface, Serializable
    private int value = 0;
    public int getValue ()
                         throws RemoteException
     return this.value;
    public void setValue (int    val)
                          throws RemoteException
     this.value = val;
    static public void main (String argv[])
     try
             // Create the object.
         RMIObject o = new RMIObject ();
             // Bind it in.
         Naming.rebind ("//localhost/RMIObject", o);
             // Now get a reference to the object from the name server.
         o = (RMIObject) Naming.lookup ("//localhost/RMIObject");
             // Set the value to 100.
             o.setValue (100);
             // Print out the value.
         System.out.println (o.getValue ());
             // Get ANOTHER reference to the object.
         o = (RMIObject) Naming.lookup ("//localhost/RMIObject");
             // See what the value is.
         System.out.println (o.getValue ());
     } catch (Exception e) {
         e.printStackTrace ();
}As you can see, I bind the object, then get a new reference to the object, set a value then get the reference again and see what the value is.
The output I get from this code is:
100
0
Which indicates that setValue is working when the reference is local but it's not updating the remote object. Now the question is, is this the way that RMI is supposed to work? i.e. should you NOT use RMI objects for holding state? If so what's the point because effectively the object would be isolated...can anyone shed any light on this?
BTW, I am NOT a student and this is NOT related to any exam/coursework question!!!

Easy folks...my reference to my potential student status was supposed to be "tongue-in-cheek", tis been a LONG time since I could get up when I want and pretend that I was really overworked and complain that I had too many assignments...
Now of course I am overworked and have too many assignments...least I get paid now...
Thanks for the help on this, I like to understand the mechanisms behind things, RMI from the outside has an appearance of "magic" that annoys me, I wasn't able to find out any description of how it actually works behind the scenes. I had to infer that it acts as a "broker", I presume that the RemoteObject object stores some information about where the JVM that actually holds the object is and then the Client uses this information (again behind the scenes) to get a connection.
This tends to worry me a bit though, the implication is that each object that is exported also contains a ServerSocket on a particular port that listens for incoming requests. Now I would presume that when it gets the request it then creates a new thread to handle it to prevent IO blocking. Now in a highly utilized system this means a lot of threads and a lot of port access, which is o.k. but that's a big use of resources. Also, if I have 2 objects listening on the same port, how does the JVM distinguish between the 2 objects when the incoming connection is received. The OS will wake up the JVM and inform it of a new connection (at least that's how Unix does it), but does the JVM have some "magic" that allows it to send it to the correct ServerSocket? I suppose that you also need to take into account Thread concurrency issues as well since you are (by definition) creating objects that will be multi-threaded...
Can anyone enlighten me on some of the above? Are there any books around that describe how RMI actually works...

Querry about Launchd behaviour

Similar Messages

Maybe you are looking for