Hello,

I have a MVBASE database currently running on Win2K server, dual CPU
system with 16G RAM.

Unfortunately, I have no documentation for this system and no one in
the company seems to know very much about how it works... (!)

The problem I have is that when CPU utilization hits 100% and stays
there for a bit, the entire database hangs and occasionally crashes the
server (so much that telnet/ssh into the windows command line does not
respond and the mouse pointer freezes)...

I wanted to know:

Is MVBASE multithreaded--meaning, is it coded specifically to take
advantage of SMP? Has anyone experienced this problem, and what do you
think it is? In my opinion based on our loads and the power of that
server, it shouldn't hang like that... is there a patch available, is
this a known issue and most importantly what can I do about it?

If I throw very powerful hardware at the problem, a brand new dual cpu
64-bit opteron system, fastest available with 16G RAM?

How can I further get to the bottom of this problem? I'd hate to buy
brand new hardware and have the situation stay the same...

thanks

picknewbie (AT) gmail (DOT) com wrote:
several patches. I'm not sure about later versions of mvBASE but
certainly if you aren't on the latest version I would consider
upgrading.

What sort of load is the machine under ? How many users?

mvBASE is normally pretty well behaved, though it can eat up processor
if it hits a tight loop in code.

I would doubt that throwing more hardware will necessarily be the
solution if you can't identify why the CPU is going to 100%. Do you
know anything that is running out of the ordinary that might be causing
this? After checking patch and version levels I would try to locate
when the 100% CPU occurs. Have you identified specifically that it is
mvBASE that is pushing the CPU to 100% and not something else?

Not much positive, just a few pointers (hopefully!)

Regards
Simon

Great, I'll look into the patches. Let me ask you this, how would I
find out what's causing the CPU load to spike so much? If there are
any suggestions at all--Windows tools or anything I can do within
MVBASE please let me know.


Also if anyone knows a good UniVerse vendor who sells it for a decent
price (meaning I can haggle down with them) I'd appreciate it. I'd
like to move to UniVerse if I can get a decent price.

thanks

simon (AT) aphroditeuk (DOT) com wrote:

The windows task manager, processes screen, has a column for CPU usage.
Click on the column header twice to sort by CPU usage. Then keep an eye on
it.


<picknewbie (AT) gmail (DOT) com> wrote

picknewbie (AT) gmail (DOT) com wrote:
is difficult looking from Windows to see what mvBASE is doing.

Depending on application, I've had some limited success in logging the
results of WHERE (V) to see whats going on at 30 second intervals (via
a program running in background that captures the result and logs to
file). Comparing the logs at the time of the CPU spike may give a
clue... It's not very scientific I'm afraid!

As far as porting away from mvBASE, we migrated to jBASE from mvBASE
without too much of a problem. OpenQM is very cost effective and
should be a good direction to migrate to as well - it is (allegedly -
Im no Universe expert!) similar to Universe.

HTH
Regards
Simon
..

Hi there,

Just a few things.

1) If you are running on Win 2K and mvBASE is running as a service you
really should be on mvBASE version 2.

2) Was the mvBASE ported from another flavour of PICK. If so I remember
there was a program to run in the SYSPROG account to update all the accounts
md entries as well as recompile and catalog all programs in DC pointed
files. I think it was something like UPGRADE was the command.

3) Also if mvBASE was ported I do remember there was an issue with some User
Exits in dictionaries that had wierd effect on the system. I remember one
that just crashed the system ( can't remember which one though )

4) Was mvBASE running on a different platform. Has this problem only arisen
since moving to new hardware and new windows.

5) Has the system been stable and has anything changed recently

6) When was the last time you did a full file Resize and restore. My system
seems to get a whole new lease of life after a good file resizing!!

I currently run a 100 user mvBASE system on an old Compaq Proliant 6500. 2 *
200MHz Pentium PRO cpu's, 2Gb memory and 18Gb of HDD
Using Microsoft NT4 server, service pack 6a
mvBASE version 1.3.02 PS9
and Cue-Bic 4GL

This system has been running for 7 years now and is still coping very well
with the daily demands of the users. I haven't archived any data off the
system ( some files have nearly a million items in them ). So you can
imagine the system takes quite a hammering when the factories ( 18 of them )
run sales reports and forecasts for the last 7 years, but the system just
keeps on trucking ( and that makes me very happy ).

Craig


<picknewbie (AT) gmail (DOT) com> wrote

Craig:

I've been looking at the performance between these older machines and the
newer servers. I'm getting huge performance improvements with the newer
machines (e.g. a 30 minute process is now completing in 2 minutes).

Secondly, if you upgrade the dbms you'd probably find that that too will
improve speed dramatically. This much speed improvement is difficult to
ignore.

Bill

"Craig" <xxxxjjjj (AT) erferf (DOT) com> wrote

Hi Bill,

Thanks for you response!

I was toying with the idea of moving up to a new server HP DL or ML with 4Gb
Memory 2 * 3.0Ghz Xeon cpu's 4 or 5 * 76Gb HDD using Raid 5

I just need to do some testing of porting our current system upto mvBASE
version2. I know it should work fine I just wanted to be sure!!

Craig

"Bill H" <you (AT) notreally (DOT) com> wrote