Skip navigation
lincoln311219
Currently Being Moderated

server instances starts automatically

May 2, 2012 7:46 AM

my server instances are starting automatically  ( I have multi server cf envronment on j-run)

 

The server log file has following logs messages

 

Primary

 

 

"Information","scheduler-20","05/01/12","10:00:41",,"Starting logging..."

"Information","scheduler-5","05/01/12","10:43:11",,"Starting logging..."

nformation","scheduler-1","05/01/12","11:37:14",,"Starting logging..."

 

 

Seconday

 

 

"Information","scheduler-20","05/01/12","10:00:55",,"Starting logging..."

 

 

The Primary service was restarted 3 times  and secondary once . Did not find much in the logs to find the reason.

 

Can anyboy have reasons why this might happen?

 
Replies
  • Currently Being Moderated
    May 2, 2012 3:17 PM   in reply to lincoln311219

    Are there any error or warning messages in the \JRun4\logs cfusion-event.log cfusion-out.log or "instance"-event and -out log?

     

    HTH, Carl.

     
    |
    Mark as:
  • Currently Being Moderated
    May 4, 2012 3:29 PM   in reply to lincoln311219

    Well, as to why a service would start automatically, that’s a built-in feature. Notice there is a jrunsvc.exe in task manager, in addition to jrun.exe. It’s job is to watch and see if a CF instance goes down and if so, it brings it back up.

     

    But as for why the service (instance) in question went down, you’d want to look at the messages in the logs BEFORE it went down, as they may give you a clue, though not always. I would recommend that rather than look in the server.log within the instance (which you can view with the CF Admin like other CF logs), you should instead look in the jrun4\logs directory and the out.log’s for your instance in question (as Carl also suggested), as they often have more valuable confirmation about overall server health.

     

    You can also see if you see messages indicating that CF was brought down (the service was “stopped”), as there would be messages before the restart about that. If there are none, then instead it may have died or been killed (from someone killing the process in task manager, typically.)

     

    Finally, there are indeed sometimes no messages at all in the out logs for why a service went down. Again, if it was killed, someone may have done that because they felt it was unresponsive. That, too, may offer no message. If you’re sure no one killed it (or stopped it), you should also look in the jrun4\bin directory to see if there are any pid.log files. They are hotspot compiler crashes, and can often offer helpful clues (though not always).

     

    Finally, if you remain stumped and need to resolve this, there are folks you can hire (myself included) who help resolve such problems by watching over your shoulders remotely while helping guide you through a process of investigation. But if you may prefer to stick with email for now, you’ll find there are many also willing to help here.

     

    /charlie arehart

     

    charlie@carehart.org

     

    Providing fast, remote, on-demand troubleshooting services for CF (and CFBuilder)

     

    More at http://www.carehart.org/consulting

     

    See also http://www.cf911.com for more on CF troubleshooting resources

     
    |
    Mark as:
  • Currently Being Moderated
    Sep 3, 2012 4:12 PM   in reply to lincoln311219

    java.lang.OutOfMemoryError: Out of swap space - Generally means some resource is full (or close to it). Sometimes hard disk that CF Java is installed on is full or even a policy may only allow limited space allocated.

    HTH, Carl.

     
    |
    Mark as:
  • Currently Being Moderated
    Sep 3, 2012 6:48 PM   in reply to carl type3

    As for OutOfMemory Swap Space errors, in my experience these are an indication rather that CF was unable to obtain the memory that it needed. (I appreciate why Carl would propose that it was related to disk space. The message surely suggests that, and indeed others on the web including Java folks will propose that’s it, but it really is not, in my experience.)

     

    Typically this is inability for CF to “get the memory it needs” is because something else is running on the server (along with CF) and using memory, so that when CF tries to use memory, it cannot. A common example is when one has a database on the same server (such as MySQL, SQL Server, etc.), or it could be a virus scanner, or the web server, or the sum of all these things.

     

    And it could be that things all “fit” fine together most of the time, but when memory used by those other things grows too high, or perhaps when CF memory use goes high at times (or at the same time), it just becomes a problem of “too many sheriffs in one town”.  

     

    Now, you can’t look at things at some one point in time (like now, as you read this and check your server’s memory use), and say “it looks like they should all fit fine right now”, because things could change later. For instance, consider that a virus scan may only use lots of memory when it runs. And if you have a database on the same server, it could use a lot more at a later time (than you see now), such as when a bunch of requests come in, and/or many large queries are run, and they push the database to load up a lot of buffers (using memory in the database address space). A burst of requests could also drive up the memory used for the web server process(es).

     

    So given the above, there are many “solutions” to this problem.

    1) You could increase the memory on the box (so more things can co-exist), which may require moving to 64-bit if you’re on 32-bit for now and maxed out in memory.

    2) You could do something to temper the other things that are using memory (move a DB server to another machine, or set it to not be able to use more than a given max memory size), or

    3) (a real “outside the box” solution to most people) you may be need to LOWER the max heap size for CF, so that it doesn’t try to use so much memory. Sometimes people set CF to use more memory than they really need. Let’s say you have a 4GB (64-bit) box and set CF to use 2GB. Maybe CF would work fine with the max heap set to 1.5GB, or even 1GB.

     

    Of course, the natural next question is “so what should I set the CF max heap size to?” No one can tell you. It depends on too many variables. But I would propose (if you can’t increase memory on the box, or lower the memory used by other things) that you should set it lower than it is now, and then watch your CF server logs (Re: server instances starts automatically\runtime\logs or Re: server instances starts automatically\logs) to watch out for any outofmemory heap space errors. Those happen when in fact the heap fills. (The “swap space”, though also an “outofmemory” error, happens in my experience BEFORE the heap fills, because of this contention.)

     

     

    Let us know if this makes sense, or what you find.

     

     

     

    /charlie

     
    |
    Mark as:
  • Currently Being Moderated
    Sep 6, 2012 1:29 PM   in reply to lincoln311219

    The key question is, did it crash with the exact same reason? Before you say yes, can you tell us how you're observing that? Which log file? And also, are you looking at just the last message in the log before it crashes? Look instead at seconds or minutes before, to see if there may be some other error as well.

     

    Second, since you're running multiserver, are you sure that the jvm.config you're changing is the one used by the instance in question? How are you confirming that? I've seen people who did not realize that they had multiple jvm.configs and different instances using different ones. I have a blog entry explaining that: http://www.carehart.org/blog/client/index.cfm/2012/6/26/identifying_wh at_instance_uses_a_given_jvm_config

     

    Please don't take offense in my asking these questions and sharing all the detail above. I can't know what you do and don't know, so am simply trying to help you (and any other readers who may find this thread in the future).

     

    Message was edited by: Charlie Arehart Corrected a couple of minor typos

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points