malachite
asked on
PB 9.0.1 Build 7171 Possible Memory Leak
I have just started a new contract at a major passenger ferry company, and their Point-of-sale system which is written in PB 9.0.1 using PFC version 6 is experiencing intermittant failures. Basically, the application is closing without warning, and with no errors. Although we have been able to replicate the problem in a lab that mimics our production environment, there is as yet no detectable pattern which causes the failure.
The system is new to me (although powerbuilder is not) and integrates with a variety of hardware (custom keyboard, custom printer, cash drawer through com port 1, etc). The team's feeling is that this is probably a memory leak.
Question 1: Is there any known memory leak issues with build 7171? Are there any major outstanding issues with upgrading from build 7171 to build 7610? I've checked the summary of all EBF's which are newer that build 7171, with no major clues as to the possible cause, and the application is so complex that it is almost impossible to get an idea of where the problem could be.
Question 2: Are there any GOOD memory leak detectors that could help determine the problem?
Question 3: Would the introduction of tracing via TraceOpen be of any help? (the thinking here is that if we could identify the last statement that was executed, this could give us a clue as to where the problem lies).
I've attached a high point value for this question, since all the developers here have been unable to resolve the issue, and my Sybase contacts have also been unable to address the issue.
Thanks for any help
Doug
The system is new to me (although powerbuilder is not) and integrates with a variety of hardware (custom keyboard, custom printer, cash drawer through com port 1, etc). The team's feeling is that this is probably a memory leak.
Question 1: Is there any known memory leak issues with build 7171? Are there any major outstanding issues with upgrading from build 7171 to build 7610? I've checked the summary of all EBF's which are newer that build 7171, with no major clues as to the possible cause, and the application is so complex that it is almost impossible to get an idea of where the problem could be.
Question 2: Are there any GOOD memory leak detectors that could help determine the problem?
Question 3: Would the introduction of tracing via TraceOpen be of any help? (the thinking here is that if we could identify the last statement that was executed, this could give us a clue as to where the problem lies).
I've attached a high point value for this question, since all the developers here have been unable to resolve the issue, and my Sybase contacts have also been unable to address the issue.
Thanks for any help
Doug
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
hi,
if its a POS system then there is no doubt that lotta third party drivers/API calls are being used to interact with the peripherals like ur pole display,cash drawer, programmable keyboard etc... So look into these calls closely...
But i still suggest an upgrade to the latest patch of 7610...
Cheers,
Rosh
if its a POS system then there is no doubt that lotta third party drivers/API calls are being used to interact with the peripherals like ur pole display,cash drawer, programmable keyboard etc... So look into these calls closely...
But i still suggest an upgrade to the latest patch of 7610...
Cheers,
Rosh
ASKER
Hi Rosh
Just FYI, today I've finally got the most up-to-date source for the system in question, and will be doing a full compile under 7610, to PBD's (which is what we use). I'll let you know the results when we've finished testing etc. Testing will take a couple of days to prove that 7610 has indeed solved or not solved the issue. If it does however not recur by the end of this week with the new build, I'll accept your answer :)
Salutations
Doug
Just FYI, today I've finally got the most up-to-date source for the system in question, and will be doing a full compile under 7610, to PBD's (which is what we use). I'll let you know the results when we've finished testing etc. Testing will take a couple of days to prove that 7610 has indeed solved or not solved the issue. If it does however not recur by the end of this week with the new build, I'll accept your answer :)
Salutations
Doug
hi,
that sounds just abt perfect...
all the best...
Cheers,
Rosh
that sounds just abt perfect...
all the best...
Cheers,
Rosh
ASKER
Yes, that's basically what Sybase said. Unfortunately, there are a LARGE number (20+) of systems that are all configured in such a way as to share the PFC and PB runtime PBD's and DLL's. A simple upgrade to 7610 is not so simple, although I am in the process now of attempting that very thing.
The shutdown occurs on all installed machines, but with seemingly no pattern (although obviously, there MUST be a pattern). Sometimes the users can go for weeks without seeing it, other times it's happening several times a shift. It's a major problem, because it forces a full reboot of the machine (not because you have to per se, but because the machines are set up in a way as to completely prevent the users from switching applications for security reasons).
I've gotten some anecdotal evidence that it may have to do with the code involved with swiping credit cards on our special keyboard (the keyboard has a built in card reader, that is used to swipe VISA, MC, etc). I'll be looking into that code on monday. I'm new to the system so I'll definitely be checking out all the code related to any special third party drivers or direct OS API calls.
Either way, I'll be recompiling the app with 7610. In your opinion, is 7610 a solid EBF build? We're on build 7171, mainly because the team is very short-staffed and didn't have the time to do upgrades etc.
Thanks for the advice
Doug