Urgent - Experiencing Run-time error '6': Overflow

[tanwill]

Hi everyone,

We are experiencing the following error message each time we try to log into the Director:

Run-time error '6':
Overflow

After which the Director just closed off completely. This problem occurred suddenly. Before that, we could log

into the Director just fine.

The DataStage version is Rel 3.6.1A running on HP-UX. I know this is a really old version, but that's how the customer wants it.
The project has been installed a few years back.

We could log into the Manager successfully and there has not been any new jobs being imported into this project

lately. we know that the jobs are running fine as the jobs were scheduled to run on a daily basis and we have

been getting results in the target tables.

Anyone has any idea as to what could have caused such an error?
Appreciate any help as this is urgent.

Thanks a lot.


Regards,
William

[ogmios]

Ai, very old version. The good news: it's a client problem.

Overflow (Error 6)
An overflow results when you try to make an assignment that exceeds the limitations of the target of the assignment. This error has the following causes and solutions:

The result of an assignment, calculation, or data type conversion is too large to be represented within the range of values allowed for that type of variable. Assign the value to a variable of a type that can hold a larger range of values.

An assignment to a property exceeds the maximum value the property can accept. Make sure your assignment fits the range for the property to which it is made.

If you did not upgrade the computer your client runs on to a new version of windows my next hunch would be that one of your original .dll got overwritten.

Ogmios

[kduke]

I doubt if you can get support on this old version. I expect you may have run out of disk space. Do a df -k and see if any are 100%. Next overflow would be from CPU memory. You might want to run top. Another option maybe a corrupted hash file. You could of hit the 2GB limit on a hash file or it was setting on a filesystem which filled up. UVFIXFILE can fix this file if it is broken. Sometimes you can clear it and rebuild it and that will fix it. Sometimes you can RESIZE HashFile * * * and that will fix it. Resize needs to create a temporary copy of the file so it can also fill up a filesystem. You may need to use a "USING" option to RESIZE. At TCL do HELP RESIZE for examples or do a search on here.

[ogmios]

Kim,

It's a client problem, not a DataStage server problem. They run the server on HP-UX.

Ogmios

[tanwill]

Hi Ogmios,

Could you please elaborate on your earlier post:
'If you did not upgrade the computer your client runs on to a new version of windows my next hunch would be that one of your original .dll got overwritten. '

How does not upgrading to a new version of windows makes an original .dll overwritten?
This 'original .dll', are you referring a .dll from windows itself or from the DataStage client?

Hi Kim,
The DataStage jobs are running fine, we have checked and the relevant file systems have not reached 100% utilization.



William

[ogmios]

If you've upgraded your client computer, you've got a completely new set of dll's and this can cause among others errors like this (new windows, old software): I'm pretty sure version 3 of DataStage is not "certified" for Windows XP e.g.

In the assumption that you did not upgrade your client computer it may be that a .dll which was used by DataStage got overwritten by a .dll from another piece of software you installed the last few days/weeks (right before you started experiencing the problems).

If you've upgraded windows you've pretty much reached the end of the life of DataStage v3 client and the only thing you can do is reinstall your previous version of Windows.
If you didn't upgrade windows what I would try to do is to reinstall the client software of DataStage and see whether that would make it work again (in the hope the correct .dlls are overwritten).
.Dll's and their versions are pretty tricky.

Ogmios

[kduke]

Ogmios, what you said makes sense. I would at least reinstall the client. See if that works.

[rasi]

Check this with another client machine. If it is the same problem revert back to your original version which it was running before and re-install Datastage Client on it.

Cheers

Rasi

[vmcburney]

You could also try creating an empty project using Administrator and seeing if Director can log into this project. This will tell you whether the problem is being caused by something specific to your existing project such as too many jobs or a corrupted repository.

[tanwill]

We tried logging in through other clients/machines but the result is the same.

We shall try to reinstall the client.

When we try to export any of the jobs from this project we encountered the same error.

However, when we logged into another project in a different server machine using the same client, there is no such error.

[tanwill]

Hi,

Just to inform everyone that the problem has been 'resolved'.
It would appear that after two of the jobs in the project has been deleted and reimported, we could log into the Director now without encountering the error.

We found that if we were to highlight these two particular jobs in the Manager before attempting to go into the Export dialog screen, the error would occurred. But this did not happened if other jobs were involved.

We still do not know what actually made these two jobs causing the Manager and Director to crash. Ascential support has suggested that certain internal numbers may have gone too high and this is why we are seeing the 'Run Time Overflow' message.

If that is the case, we may be encountering this again some time in the future when the other jobs have run the number of times to the point where the 'Overflow' may occurred again.

In the mean time, we are keeping our fingers crossed.
Thanks to everyone who reply.

William