Re: Remote debugging?

"Jeff" <Jeff_nospam_for@xxxxxxxxxxxxx> wrote in message news:mJSdnUCkbv6uvADYnZ2dnUVZ8qugnZ2d@xxxxxxxxx
Hello

I have a remote client who is experiencing crashes in an MDI application I'm responsible for. I've tried to obtain as much information as possible in the form of data files, event logs and registry settings etc as well as user descriptions but I'm still unable to recreate the crashes locally.

Is there an easy way to modify the code so that it records every event that the application receives during a given remote session and then input these locally to my version of the app and have it go through the same session and hopefully encounter the same crash?

I was wondering about overiding CWinThread::Run() and serialising the messages at that level, but I don't think this will cover messages to dialogs and other windows that might have their own threads? Is there a straightforward way to do this sort of thing or maybe there's an existing commercial product I could use?


Perhaps even if there was an automated thing like that, you couldn't repro the crash because it is due to the OS setup on the machine it is crashing on.

I would highly recommend instead you have the user start drwtsn32.exe or drwatson.exe (in C:\windows\system32) and configure it do dump minidumps. Make sure you send the client a version of your app that has been built with debug info. You don't need to send the .pdb files but you do need to send the .exe/dll's with debug info so that Dr. Watson will record stack dumps with full symbolic info.

Then when the crash happens and the .dmp file is created, have your client send it to you, then load it into Visual Studio. It will present you with a callstack with full symbolic info, and the current values of global/local variables, just as if it had crashed on your own PC. This is really neat.

-- David

.

Relevant Pages

  • Re: java.lang.OutOfMemoryError: unable to create new native thread
    ... Client can't use in crashed time. ... JVM memory is very enough and OS memory is very enough in that time. ... with runtime process. ... Some crashes are due to bugs, not the way things are designed. ...
    (comp.databases.oracle.tools)
  • Re: Recovering from a domain server crash without a backup.
    ... When a domain is created it has a unique SID. ... them and have all the client PCs sign on like nothing has changed? ... A small company with 30-50 client PCs has a single Win2K3 ... crashes and it is found that backups were not being performed. ...
    (microsoft.public.windows.server.general)
  • Re: Identifying and counting duplicates
    ... if you use client id as a row field and as a data field set to count and ... table produces is the number of clients who have had crashes. ... > That gets the data into my pivot table. ... > Client A345 had a crash which was fixed on 1/1/2005 ...
    (microsoft.public.excel.programming)
  • Re: Remote debugging?
    ... drwatson.exe and configure it do dump minidumps. ... send the .exe/dll's with debug info so that Dr. Watson will record stack ... Then when the crash happens and the .dmp file is created, have your client ... useful information from the log file I haven't yet ...
    (microsoft.public.vc.mfc)
  • Solaris telnetd failure with Heimdal client
    ... There's clearly a SEAM bug if it crashes, ... anyone know if the Heimdal client is at fault somehow, ... old) SEAM servers and all Solaris 10 updates from a month or two ago ...
    (comp.protocols.kerberos)