Sign up on the Revelation Software website to have access to the most current content, and to be able to ask questions and get answers from the Revelation community

At 05 OCT 2007 11:20:15AM Paulo Mendes wrote:

Morning,

We've been having some really weird issues every few days now, we have these batch machines that run batch jobs every morning and throughout the day, most are single jobs but sometimes we get larger jobs that get broken up into multiple threads which run individually on each batch machine as they are available, well every now and a full moon we get variable not assigned error… which could be legit problem / bug, but then when we look at all other variables being passed to the program they contain junk… I’m talking 30 characters long strings with ASCII characters from 1 - 20 mixed in with numbers and letters… and then in debugger as we keep checking each variable that gets passed into the program, eventually we get to a specific variable which causes the following:

'RTP25' Line 1. B703 Variable exceeds maximum length.

RTP25 broke at line 1 because a runtime error occurred.

RTP25 has no source code.

These machines have plenty RAM & plenty of expanded memory… We are on AREV 3.12…

Oh and if we close down the job and run-it again, same exact job… it runs and finishes just fine…

Any idea what could possibly be happening here?

Thanks

Paulo


At 05 OCT 2007 11:21AM Paulo Mendes wrote:

Oh, and all these machines are running Windows XP…


At 05 OCT 2007 12:25PM Victor Engel wrote:

In my experience, this kind of problem is network related. Sometimes it's a failing router, for example.


At 05 OCT 2007 01:29PM Paulo Mendes wrote:

Victor,

This is really strange, until like 2 weeks ago we had all dell 2000 machines, and this would still happen but only 1-2 times a month, then we moved to newer dell XP machines, and it started happening more frequent, this week alone it happened Mon, Tue and again this morning… All our TP's (Trans Processor) automatically exit and re-enter AREV sessions every hour, this way it keeps the sessions fresh… otherwise we would probably have a lot more issues than we currently have…

This is a large company and the rooms these TP machines are in get 100Mbps network connection verses the 10 we get on our workstations, so I'm not so sure it’s a network related problem… because this happens randomly between 15-20 machines…

We are currently running some tests and we’ve noticed that our memory is getting below 300 when running these jobs… which based on our past experience anything below 300 starts causing weird problems…

I'm thinking it's more of a memory issue than anything else, maybe a lot of jobs run before this specific job and by the time this job runs the session needs to be refreshed, but it’s not and it runs the job and then bombs… our TP machines also do garabagecollect & flush before each jobs it runs…

Has anyone experiences something like this before, to where variables get assigned garbage which seems to be parts and pieces object code…

Thanks

Paulo


At 05 OCT 2007 01:42PM Karen Oland wrote:

]

used to happen at novell sites with frame mismatches (using 802.3 instead of 802.2, for example) on a fairly regular basis (I'm talking several years back).

Since you replaced the machines and OS recently, I'd look at how the network is configured rather than it's hardware (although it could be a coincidence and the other hardware failed after replacing the workstations, the problem seldoms works out that way).

Also, check the hardware and memory setup on the machines - a fair number of problems can be traced to hardware in the EMS page frames (that windows lets you set up when it should not) - built in USB controllers, network controllers, video and others (on many notebooks, just the extra ROM needed can end up there). At the least, try switching to ems magic for a bit to see if it changes the symptoms (since it eliminates using hardware EMS completely).


At 05 OCT 2007 09:09PM Warren Auyong wrote:

Whenever I've seen this problem it has been due to low memory. I documented back in ARev v2.01 this type of corruption of string and variable space in low memory situations.

The next time this happens enter the "#" at the debugger prompt (!) to see a dump of available memory, string, variable and program array space.

Adding garbagecollect and flush statements strategically in the programs (usually around perform/execute statements) that blow up can often alleviate the problem - along with optimzing memory allocation. EMS Magic is recommended with XP.


At 08 OCT 2007 11:52AM Victor Engel wrote:

# doesn't dump the memory – it just prints statistics.

View this thread on the forum...

  • third_party_content/community/commentary/forums_nonworks/4aa1f59dff97abd98525736b005440bf.txt
  • Last modified: 2023/12/28 07:39
  • by 127.0.0.1