xsharp.eu

An IT company has added disks to a clients server and copied the (virtual) volumes to the new larger disks using the same drive mapping as on the old disk. After that our VO based exe started crashing when the users are working, sometimes after a few minutes, sometimes after an hour. They either get:

1 Program stops without message
2 Windows error "The program stopped working"
3 Sometimes a 5333 with In page error on i (which is logged in VOerror.log)

The IT company suspected the antivirus program and shut it down - this seems to stop the problems for half a day but today it's back with the anti virus still shut down.

The Windows Applications log shows C00000006 (which is the In Page error) many times, we think even far more than the actual crashes experienced by the users). Errors logged are on the VO Runtime DLLs, the NT DLL and the VO executable.

Does anyone have a suggestion in which direction the IT company should be searching?

Dick

Hi Dick,

Don't know if this helps or not, but just recently we had a similar problem, in our case a VO program on a particular workstation stopping at random times, normally after it had been sitting idle for some time after which Windows reported "program stopped working".

After trying loads of things (updating Windows, changing NIC, rewiring, etc) we got them to run the station in safe mode with networking, and all the problems went away. So we still don't know exactly what the problem is, but it was clearly some sort of device driver.

Nick

Dick,

Disable Data Execution Protection ?

Robert

Dick,

Some suggests that this is a hard drive error, when the OS tries to read back a memory page that was swapped to disk:

https://devblogs.microsoft.com/oldnewth ... ent-689823

Robert

Thanks for your suggestions; I forwarded those to the company. They now seem to concentrate on lost drive mappings first. The last link I sent them as well myself earlier but that appears to be a single Pc situation, not a RD environment.

If they find a solution I'll add it here for future reference.

Dick

Dick,

Try having the Users stay in the app (don't exit the app) and see if the problem gets smaller.

If so, then its probably a File Control Block issue, where Users are sharing 1 FCB, instead of having independent FCBs. Very common when using Terminal Services and other Remote Software.

Problem goes like this, the first User gets the FCB, subsequent users share it. When first User exits, other Users crash with In page errors.

Seen it dozens of times at different installs.

Paul

Seems solved now


The company responsible for the IT there has suggested to move the program from the terminal server (disk F) to the file server (disk C) while the data are still on the file server (disk F). We reinstalled ADS on the other server (disk C) and changed some path-settings and the problem seems to have gone. Or nearly gone (not sure yet).

It makes me wonder why the problem has been absent until they expanded disk space and memory...
@Paul: users don't exit the app except they were forced by the errors so that wasn't possible to test. But the FCB story made sense to me too, but how could one prevent that? Not sure if the solution I describe here basically solved that?

Dick

We have regularly the same problems. Sometimes they vanish for some months nearly completely and sometimes the re-appear in short intervals. This happens with program files/ressource files accessed by terminal server from a network path. Programs loaded completely from a local drive normally do not suffer under these problems. The reasons seems to something similar to the following:
- Terminal server cache exe/dll files with the first user accessing this file. The terminal serverreads in the program code once and have a file handle open for the resources (Dialog resource files, icons, bitmaps).
- The file handle normally remains open, also if the first user logs off.
- In certain circumstances when the user session is interrupted, the file handle gets invalid
- The terminalserver does not reload the file from the network and resources cannot be loaded anymore and the inpage error is generated.
- If a new version of the file is copied to the network directory or the file location is moved, then the terminalserver is smart enough to read in the file contents and open a new valid file handle. Otherwise only restarting the terminal server helps emptying the cache.

Arne