08-Oct-2007

Removeing the remains
After publishing the images and files of today’s walk to Trips, Tracks and Travels, the accompanying blog needed to be updated – and I found that MYSql went bazurk and was gone. That’s what triggered my attention to yesterday’s torouble with WEBES, and I found that there was still quite some pagefile in use. The reason was simply that there were still a number of WEBES programs running. After I killed them and restarted MYSql, all is pretty much in the right shape.
very much the same applied to PHP software: After Java programs were stiopped, the first invokation will give an error: “not enough core” but the next invokation will give no more error.

06-Oct-2007

Installations
New versions of software arrived, and the following have been installed:

  • FMS 2.6
  • C 7.3
  • C++ 7.3
  • PASCAL 6.0
  • FORTRAN 8.0-2
  • FORTRAN has some twists in version descrtiption, the final product has 8.0.1 on /VERSION. Would that be solved after a reboot???

    Why these updates? I’m in the process of developing some programs to replace the PHP-based blogs – simply because I get problems now and then with PHP and MYSQL since I changed the system parameters: “not enough core” at some times. This is a Unix signal with value 12 – the same as “Access Violation” on VMS? Still to figure out what exactly causes it. If it happens, it crashes the current process. No big deal, in PHP. But the MYSql Server will be gone, rendering all databases inaccessable.
    But I do have suspicions.
    The problem seems to occur when Java software is activated. Distributed Netbeans, for example, containts the Java-based IDESERVER proces, that consumes a lot of memory, even when using the Classic VM (the way I tested it). I found that PHP crashed several times – duiring the period I was busy using Distributed Netbeans…
    It can be worse, as I found out.
    WEBES
    The new software for some system management tools is web-based and is called WEBES. There are a few reasons to install it: it’s the successor of DECEvent and it might do some analysis on the error log, more elobarate than DECEvent. And it allows me to have one interface to manage all systems, because it is multi-platform: OpenVMS, Windows and Linux. For this, it is written in java: Develop once, and run it on any platform.
    That is : Almost.
    Starting the product is no big issue, it just takes time – and space: It pushes quite a lot of programs into SWAP but the amount got less once the Java programs were started. Next, I started the analysis of the error-log – and I know there is an invalid record in it.
    It never made it. that is: I have no proof at all it got anywhere, except for pushing a lot of programs out of the way, causing, for instance, MySQL to crash: “not enough core”.
    Stopping would be possible by the command
    $ DESTA STOP
    but “Desta” is unknown.
    The only way was to stop each process that seemed to belong to this group, by
    $ STOP/ID=xxxxxx

    WEBESNo, thanks