Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hello,
I'm facing an issue on putting my logs into a file. The issue is with Talend ESB, but this is a pure DI matter. ESB doesn't change anything except the number of hair I lost at this.
Context info :
[list=*]
Talend TOS 6.1.1
Windows/RHEL OS
Here is my job template :
WS_Request ==> Consumer_1==>Consumer_2==>WS_Response
Each of them have a prejob followed by a tJava :
The problem was not easy to understand at first, but I realized that System.out is a static object ... :
[list=*]
When Master Job starts, it creates a file, write some things into it and call the first consumer.
This consumer then changes the System.out, which break the log file for the master
Same thing happen for Consumer 2 which take over the System.out, and keep it for him
in the end, only the Consumer_2 can log things 😕
The scenario is an easy one, because job are ordered. In reality, PreJob is launch when the .kar is deployed on the karaf container. This was the hard thing to understand, because whenever I undeploy/deploy a WS, the order keep changing randomly.
Is there any way possible to have the console output redirected to a file, depending on jobName, and not overlapping each other.
I can't use custom component (I'm thinking about tJobInstance packages), because Webservice exports to kar/jar is broken when a component uses a library that karaf uses (like log4j).
Any help would be appreciated
My goal here is to catch anything going through the console, coming from (actually) 5 difference jobs, maybe more in the future.
I don't want to use a DB connection since it's a 24/7 up and running job.
Sincerely,
Renald.
I'm facing an issue on putting my logs into a file. The issue is with Talend ESB, but this is a pure DI matter. ESB doesn't change anything except the number of hair I lost at this.
Context info :
[list=*]
Here is my job template :
WS_Request ==> Consumer_1==>Consumer_2==>WS_Response
Each of them have a prejob followed by a tJava :
String logFile = routines.context.getLogFilePath(hostname,"MY_JOB_NAME");
System.setOut(new java.io.PrintStream(new java.io.FileOutputStream(logFile,true)));
System.setErr(new java.io.PrintStream(new java.io.FileOutputStream(logFile,true)));
The problem was not easy to understand at first, but I realized that System.out is a static object ... :
[list=*]
The scenario is an easy one, because job are ordered. In reality, PreJob is launch when the .kar is deployed on the karaf container. This was the hard thing to understand, because whenever I undeploy/deploy a WS, the order keep changing randomly.
Is there any way possible to have the console output redirected to a file, depending on jobName, and not overlapping each other.
I can't use custom component (I'm thinking about tJobInstance packages), because Webservice exports to kar/jar is broken when a component uses a library that karaf uses (like log4j).
Any help would be appreciated
My goal here is to catch anything going through the console, coming from (actually) 5 difference jobs, maybe more in the future.
I don't want to use a DB connection since it's a 24/7 up and running job.
Sincerely,
Renald.
Hello shong,
I've used jobName in my tJava to create the filename, that part works, but the problem is that System.out is unique (because it's static). So if you declare another file, System.out will just move to this file because it's in one unique JVM.
This is my front job structure :
My goal is to have 3 WS, all in "keep listening" mode :
[list=*]
FRONT (the example above)
CONSUMER_1
CONSUMER_2
And so 3 separate log, with timestamps in milliseconds. In need to be able to say :
[list=*]
This part took XX millis
The write of the file took X milli
the whole job took XX milli
The front job, the one in example, is in keep listening mode, which cannot be used with tStatCatcher and tLogCatcher
(this stuff is pretty awesome for what I need, but I must pass sadly).
I've looked at tRedirectOutput, and it's exactly the same code inside the component, compared to the one I typed above. Also, there's the problem of packaging the bundle with custom component, so it might not be a good idea to use one.
My colleague worked with me on the subject (and confirmed that the building of the bundle doesn't take external libs since it conflict with native libraries). We tried to alter the custom component to completely deactivate the logging part, and the problem is still there.
So we went on a more "hard coded" logging solution : modifying lo4j conf like that (the first answer, not the original post) :
Is it worth it to throw some hours in that subject, considering the fact that is quite stable, but could be changed without hours of rework if I want a different separation of my files ?
Thx in advance for your help.
I've used jobName in my tJava to create the filename, that part works, but the problem is that System.out is unique (because it's static). So if you declare another file, System.out will just move to this file because it's in one unique JVM.
This is my front job structure :
My goal is to have 3 WS, all in "keep listening" mode :
[list=*]
And so 3 separate log, with timestamps in milliseconds. In need to be able to say :
[list=*]
The front job, the one in example, is in keep listening mode, which cannot be used with tStatCatcher and tLogCatcher
(this stuff is pretty awesome for what I need, but I must pass sadly).
I've looked at tRedirectOutput, and it's exactly the same code inside the component, compared to the one I typed above. Also, there's the problem of packaging the bundle with custom component, so it might not be a good idea to use one.
My colleague worked with me on the subject (and confirmed that the building of the bundle doesn't take external libs since it conflict with native libraries). We tried to alter the custom component to completely deactivate the logging part, and the problem is still there.
So we went on a more "hard coded" logging solution : modifying lo4j conf like that (the first answer, not the original post) :
Is it worth it to throw some hours in that subject, considering the fact that is quite stable, but could be changed without hours of rework if I want a different separation of my files ?
Thx in advance for your help.