Java high memory question - OS2

This is a discussion on Java high memory question - OS2 ; Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically use high memory? And thus can expand the total heap size? If not, can one use the CONFIG.SYS file for either or both of these Java toolsets to do ...

+ Reply to Thread
Results 1 to 17 of 17

Thread: Java high memory question

  1. Java high memory question

    Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically use high
    memory? And thus can expand the total heap size?

    If not, can one use the CONFIG.SYS file for either or both of these Java
    toolsets to do this as in:

    SET JAVA_HIGH_MEMORY=1

    The reason I ask is that the major heap full errors one sees in Norman's NVC
    5.8 for OS/2, if Java 1.1.8 or Java 1.4.2 are not running in high memory sure
    might be cured by this. Knowing also that NVC is documented as not being
    compatible with IBM's Java 1.1.3 as I've read.

    Thoughts?


    --


    --> Sleep well; OS2's still awake!

    Mike Luther

  2. Re: Java high memory question

    Hi Mike

    Mike Luther wrote:
    > Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically
    > use high memory? And thus can expand the total heap size?
    >
    > If not, can one use the CONFIG.SYS file for either or both of these Java
    > toolsets to do this as in:
    >
    > SET JAVA_HIGH_MEMORY=1
    >



    I have the above line in my config.sys file; no idea if it actually
    works with java142...

    Regards

    Pete



    > The reason I ask is that the major heap full errors one sees in Norman's
    > NVC 5.8 for OS/2, if Java 1.1.8 or Java 1.4.2 are not running in high
    > memory sure might be cured by this. Knowing also that NVC is documented
    > as not being compatible with IBM's Java 1.1.3 as I've read.
    >
    > Thoughts?
    >
    >


  3. Re: Java high memory question

    Thanks Peter

    Peter Brown wrote:
    > Hi Mike
    >
    > Mike Luther wrote:
    >> Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically
    >> use high memory? And thus can expand the total heap size?
    >>
    >> If not, can one use the CONFIG.SYS file for either or both of these
    >> Java toolsets to do this as in:
    >>
    >> SET JAVA_HIGH_MEMORY=1
    >>

    >
    >
    > I have the above line in my config.sys file; no idea if it actually
    > works with java142...
    >
    > Regards
    >
    > Pete
    >
    >
    >
    >> The reason I ask is that the major heap full errors one sees in
    >> Norman's NVC 5.8 for OS/2, if Java 1.1.8 or Java 1.4.2 are not running
    >> in high memory sure might be cured by this. Knowing also that NVC is
    >> documented as not being compatible with IBM's Java 1.1.3 as I've read.
    >>
    >> Thoughts?
    >>
    >>


    Thanks Peter. From what you just posted, I can then guess that your system is
    actually working on a 'default' of VIRTUALMEMORYADDRESS=2048, which I think I
    understand is the value which is used if you use what you posted and do not
    specify otherwise for the value. Also from what I read, there are people using
    SNAP, which I use, which have had system lockups during the boot run unless the
    value is BELOW the 2048 amount. I think I've seen more than one person suggest
    that the value be set at 1356, can't remember the exact amount from memory and
    where I am typing this.

    Any other help would be welcome here...


    --


    --> Sleep well; OS2's still awake!

    Mike Luther

  4. Re: Java high memory question

    On 12/06/07 08:13 am, Mike Luther wrote:
    > Thanks Peter
    >
    > Peter Brown wrote:
    >> Hi Mike
    >>
    >> Mike Luther wrote:
    >>> Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically
    >>> use high memory? And thus can expand the total heap size?
    >>>
    >>> If not, can one use the CONFIG.SYS file for either or both of these
    >>> Java toolsets to do this as in:
    >>>
    >>> SET JAVA_HIGH_MEMORY=1
    >>>

    >>
    >>
    >> I have the above line in my config.sys file; no idea if it actually
    >> works with java142...
    >>
    >> Regards
    >>
    >> Pete
    >>
    >>
    >>
    >>> The reason I ask is that the major heap full errors one sees in
    >>> Norman's NVC 5.8 for OS/2, if Java 1.1.8 or Java 1.4.2 are not
    >>> running in high memory sure might be cured by this. Knowing also that
    >>> NVC is documented as not being compatible with IBM's Java 1.1.3 as
    >>> I've read.
    >>>
    >>> Thoughts?
    >>>
    >>>

    >
    > Thanks Peter. From what you just posted, I can then guess that your
    > system is actually working on a 'default' of VIRTUALMEMORYADDRESS=2048,
    > which I think I understand is the value which is used if you use what
    > you posted and do not specify otherwise for the value. Also from what I
    > read, there are people using SNAP, which I use, which have had system
    > lockups during the boot run unless the value is BELOW the 2048 amount. I
    > think I've seen more than one person suggest that the value be set at
    > 1356, can't remember the exact amount from memory and where I am typing
    > this.
    >
    > Any other help would be welcome here...
    >
    >


    I just fired up an Java app with Innotek Java and looked at the memory
    with Theseus. Innowin had allocated a couple of chunks in high memory. I
    do have set JAVA_HIGH_MEMORY in config.sys but don't think it applies to
    1.4. It is safer to have VIRTUALMEMORYADDRESS= lower than 2048 but it
    really depends on your system.
    You could experiment
    Dave


  5. Re: Java high memory question

    Sir:

    Mike Luther wrote:
    > Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically
    > use high memory? And thus can expand the total heap size?

    Yes to both, but only if authorized by the following statement:
    >
    > SET JAVA_HIGH_MEMORY=1
    >

    Available only on WSeB and Post fp 12 OS/2. See the Readme with Java 1.1.8.
    > The reason I ask is that the major heap full errors one sees in Norman's
    > NVC 5.8 for OS/2, if Java 1.1.8 or Java 1.4.2 are not running in high
    > memory sure might be cured by this. Knowing also that NVC is documented
    > as not being compatible with IBM's Java 1.1.3 as I've read.
    >
    > Thoughts?
    >
    >

    VIRTUALADDRESSLIMIT=2048 is the default in the post fp12 systems.
    The lower limit that some find cure SNAP problems (when they have too
    much memory on their video card and don't limit it to only the needed
    frame buffer size.) is 1536, one GiB above the 512MiB pre fp13 default
    limit. I'd suppose that when video cards appear with 2 GiB of video
    memory, they'll have to drop that down more.

    --
    Bill
    Thanks a Million!

  6. Re: Java high memory question

    On 12/06/07 07:11 pm, William L. Hartzell wrote:

    > VIRTUALADDRESSLIMIT=2048 is the default in the post fp12 systems.


    No, 1024 is the default.

    > The lower limit that some find cure SNAP problems (when they have too
    > much memory on their video card and don't limit it to only the needed
    > frame buffer size.)


    There is no SNAP problem, it's a memory addressing problem. There's
    only 4GB of address space, and OS/2 does not dynamically adjust the
    different pools it uses. If the video frame buffer takes a big chunk,
    that limits the amount that can be used for other things. IBM used to
    limit the frame buffer to 32MB, but lots of people complained. There's
    more info here:
    http://www.scitechsoft.com/support/faq/fom.cgi?file=21
    http://www.scitechsoft.com/support/faq/fom.cgi?file=33

  7. Re: Java high memory question

    Precious information for me Steve

    Steve Wendt wrote:
    > On 12/06/07 07:11 pm, William L. Hartzell wrote:
    >
    >> VIRTUALADDRESSLIMIT=2048 is the default in the post fp12 systems.

    >
    > No, 1024 is the default.
    >


    AHA! That may be why when I tested this today trying to clean up HEAP FULL
    errors in total scans with NORMAN 5.8 latest everything, I got no change for a
    CONFIG.SYS setting of VIRTUALADDRESSLIMIT=1024.

    Later today I'll reset a test run to 2048 and see if that takes out at least
    some of the HEAP FULL errors.

    >> The lower limit that some find cure SNAP problems (when they have too
    >> much memory on their video card and don't limit it to only the needed
    >> frame buffer size.)

    >
    > There is no SNAP problem, it's a memory addressing problem. There's
    > only 4GB of address space, and OS/2 does not dynamically adjust the
    > different pools it uses. If the video frame buffer takes a big chunk,
    > that limits the amount that can be used for other things. IBM used to
    > limit the frame buffer to 32MB, but lots of people complained. There's
    > more info here:
    > http://www.scitechsoft.com/support/faq/fom.cgi?file=21
    > http://www.scitechsoft.com/support/faq/fom.cgi?file=33


    I'll be able to take time later here to read these references. Study this trap
    process and scenario which I really didn't think about. Thanks for starting me
    down this learning journey.


    --


    --> Sleep well; OS2's still awake!

    Mike Luther

  8. Re: Java high memory question

    Hi Mike

    Mike Luther wrote:
    > Thanks Peter
    >
    > Peter Brown wrote:
    >> Hi Mike
    >>
    >> Mike Luther wrote:
    >>> Does Java 1.1.8 and/or Java 1.4.2 from Innotek for OS/2 automatically
    >>> use high memory? And thus can expand the total heap size?
    >>>
    >>> If not, can one use the CONFIG.SYS file for either or both of these
    >>> Java toolsets to do this as in:
    >>>
    >>> SET JAVA_HIGH_MEMORY=1
    >>>

    >>
    >>
    >> I have the above line in my config.sys file; no idea if it actually
    >> works with java142...
    >>
    >> Regards
    >>
    >> Pete
    >>
    >>
    >>
    >>> The reason I ask is that the major heap full errors one sees in
    >>> Norman's NVC 5.8 for OS/2, if Java 1.1.8 or Java 1.4.2 are not
    >>> running in high memory sure might be cured by this. Knowing also
    >>> that NVC is documented as not being compatible with IBM's Java 1.1.3
    >>> as I've read.
    >>>
    >>> Thoughts?
    >>>
    >>>

    >
    > Thanks Peter. From what you just posted, I can then guess that your
    > system is actually working on a 'default' of VIRTUALMEMORYADDRESS=2048,



    No, I do not have a VIRTUALMEMORYADDRESS line in config.sys

    Now I guess I ought to check if I should...

    No mention of that setting in Config Tool.


    Maybe you are looking at this value

    VIRTUALADDRESSLIMIT=1536

    which is the value I use.

    > which I think I understand is the value which is used if you use what
    > you posted and do not specify otherwise for the value. Also from what I
    > read, there are people using SNAP, which I use, which have had system
    > lockups during the boot run unless the value is BELOW the 2048 amount.



    I'm using SNAP on an ACPI 2.22 system and do not have any problems
    booting usually - Shutdown is a different matter though where the only
    thing that usually works is the Power Off option.


    > I think I've seen more than one person suggest that the value be set at
    > 1356, can't remember the exact amount from memory and where I am typing
    > this.
    >
    > Any other help would be welcome here...
    >
    >




    Regards

    Pete

  9. Re: Java high memory question

    On Fri, 7 Dec 2007 14:42:44 UTC, Mike Luther
    wrote:

    > AHA! That may be why when I tested this today trying to clean up HEAP FULL
    > errors in total scans with NORMAN 5.8 latest everything, I got no change for a
    > CONFIG.SYS setting of VIRTUALADDRESSLIMIT=1024.


    There is only ONE cure for HEAP FULL, in NVC. Quit using it. There is
    a basic problem with the program, and NORMAN has declined to develop
    the OS/2 version further, to fix problems. I have had to quit using
    NVC altogether because of that sort of nonsense.
    VIRTUALADDRESSLIMIT=1536 seems to be optimal, but it doesn't fix
    anything, it simply makes the problem happen less often. I have
    installed ClamAV in it's place, but the documentation abbout how to
    actually use ClamAV is pathertic, at best (typical *NIX crap, with too
    many, totally useless, command line parameters, most of which don't
    make sense). On the other hand, it does seem to actually work, once
    you figure it out, although I have, yet, to see it trap a virus (my
    ISP filters those things out, before they get to me).

    Briefly, what I did, was:
    Get CalmAV from HOBBES:

    > http://hobbes.nmsu.edu/pub/os2/util/...0_91_2-os2.wpi


    Install it.
    Make a shadow of the ClamAV daemon in the Startup folder.
    Make program icon for Freshclam:
    Path and file name = \Freshclam.exe
    Parameters = --config-file=../etc/freshclamdb.conf
    Working directory =

    I then copied ..\etc\freshclam.conf to ..\freshclam\freshclamdb.conf
    (Note the forward slashes in the parameters field). Then, I edited
    freshclamdb.conf to make a log file, and reflect the proper
    DatabaseMirror server for my country, and I turned on the NotifyClamd
    setting. Then, I use the scheduler feature of DragText, to run that
    every 4 hours (set the time so it is not on the hour, or half hour,
    where a LOT of people put it). Now, when Freshclam runs, it should
    download updates, abd tell the daemon that they are ready. The daemon,
    in turn, will load up the new information.

    Okay, now we have the base running. Chances are, that you will want to
    manually run a scan on drives, folders, or files, on demand. Make
    another program icon, for ClamDscan:
    Path and file name = \ClamDscan.exe
    Parameters = -v --config-file=../etc/clamddb.conf -l
    ../Logs/ClanDscan.log -m
    Working directory =

    Again, note the forward slashes. In the Session tab, turn OFF, Close
    window on exit (so you can see the results, before closing he window).

    Now, comes the tricky part. First, find the Text Editor icon (usually
    in Programs-> Utilities), add an association to that icon, with "*"
    (without the quotes) That will make the Text Editor be the an
    associated program for EVERY file (I know that it is already the
    default, but when you do the next step, that gets broken). Now, do the
    same association with the ClamDscan icon, which adds that to the list.
    Now, if you right mouse click on ANY file icon, and select the arrow
    beside Open As, one of the options is ClamDscan. It should not be the
    default. use AssocEdit to fix that problem, if it happens.

    Okay, that does individual files, but you also want to be able to do
    folders, and whole drives. This trick works in eCS (at least in v 1.2,
    and up). Find the Extended Menu options folder. That should be in
    System Setup, or System Setp-> Appearance. Then, make a shadow of the
    ClamDscan program in there. Now, when you right mouse click on a
    drive, or folder, one of the options is ClamDscan.

    Hope that helps...
    --
    From the eComStation 2.0 RC2 of Doug Bissett
    dougb007 at telus dot net
    (Please make the obvious changes, to e-mail me)


  10. Re: Java high memory question

    On Fri, 07 Dec 2007 23:53:31 GMT, Doug Bissett wrote:

    > I have installed ClamAV in it's place, but the documentation abbout how
    > to actually use ClamAV is pathertic, at best (typical *NIX crap, with too
    > many, totally useless, command line parameters, most of which don't
    > make sense). On the other hand, it does seem to actually work, once
    > you figure it out, although I have, yet, to see it trap a virus (my
    > ISP filters those things out, before they get to me).


    How do you know it works if you've never had a virus to trap and what is
    the point in the first place if they're filtered out before they get to you?
    The things some people do just beggars belief sometimes.

  11. Re: Java high memory question

    Sir:

    Paul Ratcliffe wrote:
    > On Fri, 07 Dec 2007 23:53:31 GMT, Doug Bissett wrote:
    >
    >> I have installed ClamAV in it's place, but the documentation abbout how
    >> to actually use ClamAV is pathertic, at best (typical *NIX crap, with too
    >> many, totally useless, command line parameters, most of which don't
    >> make sense). On the other hand, it does seem to actually work, once
    >> you figure it out, although I have, yet, to see it trap a virus (my
    >> ISP filters those things out, before they get to me).

    >
    > How do you know it works if you've never had a virus to trap and what is
    > the point in the first place if they're filtered out before they get to you?
    > The things some people do just beggars belief sometimes.

    I have a folder that contains several e-mails from an associate which do
    have virus attached. ClamAV happily finds all of them. Ya, these virus
    are ancient so much that even IBM AV finds them. But they make a valid
    test case.

    As to possessing an AV product when one's ISP does filter? They will
    drop lessor seen virus signatures from their filters to keep them fast.
    All we need is a blast from the past to know that not all filters are
    equal. One of those appeared recently and Dr Solomon had to readd it to
    their filters according to an article I saw on the Register last September.
    --
    Bill
    Thanks a Million!

  12. Re: Java high memory question

    First experiment done Dave

    Dave Yeo wrote:

    > I just fired up an Java app with Innotek Java and looked at the memory
    > with Theseus. Innowin had allocated a couple of chunks in high memory. I
    > do have set JAVA_HIGH_MEMORY in config.sys but don't think it applies to
    > 1.4. It is safer to have VIRTUALMEMORYADDRESS= lower than 2048 but it
    > really depends on your system.
    > You could experiment
    > Dave
    >


    Post from the experiment R40. There is absolutely no difference between using
    CONFIG.SYS with no mention of Java High Memory and Virtual memory and setting
    these into it, either with a memory limit of 1024 or 1536 here. I get exactly
    the same pile of HEAP FULL errors in a full scan of this roughly 300,000 file
    laptop either way. In this case, what I had to do do even get the Norman scan
    to complete was to selectively go through and add each HEAP FULL error starting
    with the boot partition C: from the top down, then continuing on that way on D:
    and E: until I stopped getting SYS3171 errors like:

    >
    > 12-06-2007 02:59:05 SYS3171 PID 008e TID 0002 Slot 0074
    > C:\NORMAN\NVC\BIN\NVCOD.EXE
    > c0000005
    > 000270f6
    > P1=00000001 P2=8304438f P3=XXXXXXXX P4=XXXXXXXX
    > EAX=2ecf0000 EBX=000517b7 ECX=0000000e EDX=00000007
    > ESI=00000000 EDI=000517b7
    > DS=0053 DSACC=f0f3 DSLIM=ffffffff
    > ES=0053 ESACC=f0f3 ESLIM=ffffffff
    > FS=150b FSACC=00f3 FSLIM=00000030
    > GS=0000 GSACC=**** GSLIM=********
    > CS:EIP=005b:1c021d33 CSACC=f0df CSLIM=ffffffff
    > SS:ESP=0053:05d90000 SSACC=f0f3 SSLIM=ffffffff
    > EBP=05d90068 FLG=00012286
    >
    > NVCOD.EXE 0002:000070f6


    That just to get to the point where the complete hard drive scan would finish,
    but yet with I think nine remaining HEAP FULL errors.

    The point about citing the SYS3171 errors is that as far as my research goes,
    there seems to be more than just a HEAP FULL error problem here. Yes, the
    error posts in the Norman scan. However, as shown, further on down the way,
    even with no file involved that is another HEAP FULL error, we get memory
    corruption during the scan from a file which isn't involved in that.

    Further, if I start the scan from a drive other than C:, the entire HEAP FULL
    and future SYS3171 risk, plus what I seem to see is other memory corruption
    issues in the system from later use after the scan, change the pattern of failure.

    I posted all this to the Norman forum and corresponded with the Developer long
    before the turn off of development with Norman. Indeed, per my notes, there
    was improvement in the issue made. Adding the /EX:65535 command line change in
    CONFIG.SYS did help on this R40, for example.

    But whatever is at issue here must really be a memory bear big time that is
    hard to spot in the forest. As well, I don't have the training to track this
    in Theseus, nor without the source, could ever put a flag file step by step
    trace to determine precisely what file action to stop the step-by-step action,
    so I could look at it with Theseus and really fix it, if possible.


    Mike Luther

  13. Re: Java high memory question

    Thanks Doug. You work hard to help us all ..

    Doug Bissett wrote:

    > There is only ONE cure for HEAP FULL, in NVC. Quit using it. There is
    > a basic problem with the program, and NORMAN has declined to develop
    > the OS/2 version further, to fix problems.



    I understand that Norman didn't get it fixed. The problem is that in my case,
    I see reasons that there seems to be no other tool that will address some of
    what I think I need that it does.

    > VIRTUALADDRESSLIMIT=1536 seems to be optimal, but it doesn't fix
    > anything, it simply makes the problem happen less often.


    I've now tested this R40 with a limit of 1024 and 1536. There is no change at
    all in the HEAP FULL roll out to the end of the scan. Now that I have a bunch
    of them already excluded in the exclude file setup on this box.

    > I have
    > installed ClamAV in it's place, but the documentation abbout how to
    > actually use ClamAV is pathertic, at best (typical *NIX crap, with too
    > many, totally useless, command line parameters, most of which don't
    > make sense). On the other hand, it does seem to actually work, once
    > you figure it out, although I have, yet, to see it trap a virus (my
    > ISP filters those things out, before they get to me).


    In this case, when Suddenlink bought out COX and became my IP provider here,
    they completely changed at least the Email operations to include virus checking
    operations. As well, per what I think is correct, they are still a full UNIX
    operation, and not Windows oriented for their core operations. At that point I
    saw a HUGE reduction in virus and spam traffic in my Email operations.


    But that doesn't mean that I can afford to not think about things like java
    issues. Since that time, just before the Mozilla crew 'fixed' known java
    security issues with one of the Seamonkey releases for OS/2 I actually got to
    see two of the on-access site deals, somehow, where my proxy settings for the
    Seamonkey setup for that were changed remotely!!

    I use Privoxy 3+ latest for SMK. I have no idea what site or action did this,
    but the proxy settings in it were changed remotely, in line with exactly what
    the crew said was a possible security issue for java with Seamonkey. In this
    case, I caught it because the entire proxy server operation I normally used
    stopped working the next time I fired up Seamonkey. OK, I fixed it.

    But I guarantee you, just because you run OS/2 doesn't mean you are free from
    virus and evil operations. Any cross platform common operations code process
    has the ability to be used for evil purposes. Even if it is DOS.

    > Briefly, what I did, was:
    > Get CalmAV from HOBBES:


    Remainder printed for further study and education.

    > Hope that helps...


    Yes. I've not moved to testing ClamAV but it is on the radar, sigh ..


    Mike Luther

  14. Re: Java high memory question

    You too, Bill, are a treasure of help for all of us here..

    William L. Hartzell wrote:

    > As to possessing an AV product when one's ISP does filter? They will
    > drop lessor seen virus signatures from their filters to keep them fast.
    > All we need is a blast from the past to know that not all filters are
    > equal.


    In my experience, yes, I agree this does happen. Not yet with my current
    Suddenlink provider, but earlier elsewhere.

    As well, there is always the problem that an attachment contains a problem
    which is there that isn't even known yet! Even for months into the future.
    Then when it 'escapes' you still need the ability to scan for even the old
    stuff to expose what wasn't known in the past.

    Just my opinion.


    Mike Luther

  15. Re: Java high memory question

    On 12/09/07 06:10 am, Mike Luther wrote:
    > First experiment done Dave
    >
    > Dave Yeo wrote:
    >
    >> I just fired up an Java app with Innotek Java and looked at the memory
    >> with Theseus. Innowin had allocated a couple of chunks in high memory.
    >> I do have set JAVA_HIGH_MEMORY in config.sys but don't think it
    >> applies to 1.4. It is safer to have VIRTUALMEMORYADDRESS= lower than
    >> 2048 but it really depends on your system.
    >> You could experiment
    >> Dave
    >>

    >
    > Post from the experiment R40. There is absolutely no difference between
    > using CONFIG.SYS with no mention of Java High Memory and Virtual memory
    > and setting these into it, either with a memory limit of 1024 or 1536
    > here. I get exactly the same pile of HEAP FULL errors in a full scan of
    > this roughly 300,000 file laptop either way. In this case, what I had to
    > do do even get the Norman scan to complete was to selectively go through
    > and add each HEAP FULL error starting with the boot partition C: from
    > the top down, then continuing on that way on D: and E: until I stopped
    > getting SYS3171 errors like:
    >
    >>
    >> 12-06-2007 02:59:05 SYS3171 PID 008e TID 0002 Slot 0074
    >> C:\NORMAN\NVC\BIN\NVCOD.EXE
    >> c0000005
    >> 000270f6
    >> P1=00000001 P2=8304438f P3=XXXXXXXX P4=XXXXXXXX
    >> EAX=2ecf0000 EBX=000517b7 ECX=0000000e EDX=00000007
    >> ESI=00000000 EDI=000517b7
    >> DS=0053 DSACC=f0f3 DSLIM=ffffffff
    >> ES=0053 ESACC=f0f3 ESLIM=ffffffff
    >> FS=150b FSACC=00f3 FSLIM=00000030
    >> GS=0000 GSACC=**** GSLIM=********
    >> CS:EIP=005b:1c021d33 CSACC=f0df CSLIM=ffffffff
    >> SS:ESP=0053:05d90000 SSACC=f0f3 SSLIM=ffffffff
    >> EBP=05d90068 FLG=00012286
    >>
    >> NVCOD.EXE 0002:000070f6

    >
    > That just to get to the point where the complete hard drive scan would
    > finish, but yet with I think nine remaining HEAP FULL errors.
    >
    > The point about citing the SYS3171 errors is that as far as my research
    > goes, there seems to be more than just a HEAP FULL error problem here.
    > Yes, the error posts in the Norman scan. However, as shown, further on
    > down the way, even with no file involved that is another HEAP FULL
    > error, we get memory corruption during the scan from a file which isn't
    > involved in that.
    >
    > Further, if I start the scan from a drive other than C:, the entire HEAP
    > FULL and future SYS3171 risk, plus what I seem to see is other memory
    > corruption issues in the system from later use after the scan, change
    > the pattern of failure.
    >
    > I posted all this to the Norman forum and corresponded with the
    > Developer long before the turn off of development with Norman. Indeed,
    > per my notes, there was improvement in the issue made. Adding the
    > /EX:65535 command line change in CONFIG.SYS did help on this R40, for
    > example.
    >
    > But whatever is at issue here must really be a memory bear big time that
    > is hard to spot in the forest. As well, I don't have the training to
    > track this in Theseus, nor without the source, could ever put a flag
    > file step by step trace to determine precisely what file action to stop
    > the step-by-step action, so I could look at it with Theseus and really
    > fix it, if possible.
    >
    >
    > Mike Luther


    I'd say at this point Norman just can't handle todays large drives/large
    number of files. As others have mentioned clamav seems to work. At least
    here it has identified viruses I have laying around and new ones that
    have come in with spam.
    The other question is does OS/2 even need to be scanned for viruses?
    There are pretty well no OS/2 specific viruses and the number of DOS
    viruses out there must be close to zero now.
    The only virus I ever had was a boot sector virus that came on a floppy
    that came with a broken IDE controller back when I got my first CDrom.
    It did actually infect OS/2 DOS sessions and surprisingly it was WINOS2
    that first spotted it. IIRC DOS virus scanner also spotted it and I just
    did a fdisk /newmbr to cure it.
    Of course if you are trading files with Win systems then a virus scan
    can help protect them
    Dave


  16. Re: Java high memory question

    On Sun, 9 Dec 2007 19:07:14 UTC in comp.os.os2.apps, Dave Yeo
    wrote:

    > As others have mentioned clamav seems to work.


    I see no reason why Clamav should work any less well on OS/2 than it does on
    other platforms. I run a mailserver for the company that I work for and all
    inbound mail goes though ClamAV and SpamAssassin. We haven't had any viruses
    escape it since I implemented it about 5 months ago. In the last month it has
    intercepted 785 mail bourne viruses and phishing attempts. We then have Sophos
    on all the desktops and it has never had to catch a single mail virus - they've
    already been cleaned by ClamAV.

    --
    Trevor Hemsley, Brighton, UK
    Trevor dot Hemsley at ntlworld dot com

  17. Re: Java high memory question

    On Sat, 8 Dec 2007 02:18:03 UTC, Paul Ratcliffe
    wrote:

    > On Fri, 07 Dec 2007 23:53:31 GMT, Doug Bissett wrote:
    >
    > > I have installed ClamAV in it's place, but the documentation abbout how
    > > to actually use ClamAV is pathertic, at best (typical *NIX crap, with too
    > > many, totally useless, command line parameters, most of which don't
    > > make sense). On the other hand, it does seem to actually work, once
    > > you figure it out, although I have, yet, to see it trap a virus (my
    > > ISP filters those things out, before they get to me).

    >
    > How do you know it works if you've never had a virus to trap and what is
    > the point in the first place if they're filtered out before they get to you?


    I don't trust the windows stuff, that my ISP uses...

    > The things some people do just beggars belief sometimes.


    It does, doesn't it?

    May you never regret not protecting yourself...
    --
    From the eComStation 2.0 RC2 of Doug Bissett
    dougb007 at telus dot net
    (Please make the obvious changes, to e-mail me)


+ Reply to Thread