I am using Simon Green's NHibernate Helper Project http://blogs.intesoft.net/simon/articles/16.aspx (Thanks Simon) to manage sessions, Forms Authentication and In Proc mode to save some objects in the Session. Sometimes it works fine, but often the application just ends with no warning.
I cannot replicate this behavior by following the same sequence of events. Also, on my computer and our company

If this is an ASP.NET application perhaps you've got a memory leak somewhere? ASP.NET will automatically kill and restart the ASP.NET worker process if it grows beyond 60% of available memory.

Just a thought.

Cheers,

Symon.

wouldn't I be able to replicate the error consistently? If I open the browser and go to exactly the same pages in the same order wouldn't the amount of memory used be about the same? Also, this didn't used to be a problem until I made changes to the session handling in NHibernate.

I solved this issue in an older app once by dissabling the virus scanner on the machine this was happening on. Apparently virus scanners 'touch' the Web.configs and bin files making the server restart.

I currently have the same inssue in another application and cannot solve.

I turned all logging off and the problem went away. After a bit of testing, I found that logging lots of info (levels INFO and DEBUG) caused the appliction to suddenly end but some logging (WARN,DEBUG) seem fine. Does anyone have an explaination for this? I tried the log4net-user mailing list but haven't gotten any answers yet.

Here is a bit more information on the logging I was doing:
The logging files were being locked so I upgraded to a later version of log4net (1.2.9) to use FileAppender+MinimalLock. This only locks the file while it is being written to and then releases it. And in my web config file I have:
<runtime>
<assemblyBinding xmlns="urn:schemas-microsoft-com:asm.v1">
<dependentAssembly>
<assemblyIdentity name="log4net"
publicKeyToken="b32731d11ce58905"
culture="neutral" />
<bindingRedirect oldVersion="1.2.0.30714"
newVersion="1.2.9.0"/>
</dependentAssembly>
</assemblyBinding>
</runtime>

Hmm, could this be caused by a NullReferenceException while building a log message? By the way, NH has been using log4net 1.2.9 for a long time now.

also happen on my machine and our company's server - not just on the production server? Also the problem wouldn't happen intermittently would it?

I think there was one case where logging on Flush caused an NRE, depending on which entities were present in the session, or something like that. Which version of NH are you using, by the way?

I am using version 0.5.0.0 - I guess it is time to upgrade...