.NET Remoting connection timing out after many invocations
-
Hi guys, So I have an object that I'm exposing via .NET remoting (for historical / political reasons I can't just drop this and go to something more modern), which I use to process a list of items, one call per item in my list. The remoting server is set up as Single call rather than Singleton, and so my understanding is that I should get a new instance of the server class on each invocation - i.e. for each item in my list. This seems to work just fine for small datasets, but when I push it a little by processing a list containing approx. 200K items, it reaches around 19K processed, I see a timeout, and the call fails with a socket exception - "A connection attempt failed because the connected party did not properly respond after a period of time or established connection failed because connected host has failed to respond". The next call then succeeds and processing continues...but then fails again later. There seems to be no significance in the number of records processed - this varies from around 19K to 19.1k and fails at different points and with different records. Additionally, when I run the system in multi-threaded mode (Parallel.ForEach() & MaxDegreeOfParallelism == -1), then I'm seeing the same "profile" of failures, but instead of a single record fail followed by the next "working", I'm seeing about 10 failures before it comes back to life - which leads me to believe it's something like the server actually dying and being re-instantiated. Anyone have any ideas? I can batch out the data and get a "new" remoting connection if I create a factory to do so (I'm injecting my remoting client), but this seems like avoiding the problem rather than understanding the cause and fixing it. Ultimately this will be used to process > 1.2M records Cheers guys!
C# has already designed away most of the tedium of C++.
-
Hi guys, So I have an object that I'm exposing via .NET remoting (for historical / political reasons I can't just drop this and go to something more modern), which I use to process a list of items, one call per item in my list. The remoting server is set up as Single call rather than Singleton, and so my understanding is that I should get a new instance of the server class on each invocation - i.e. for each item in my list. This seems to work just fine for small datasets, but when I push it a little by processing a list containing approx. 200K items, it reaches around 19K processed, I see a timeout, and the call fails with a socket exception - "A connection attempt failed because the connected party did not properly respond after a period of time or established connection failed because connected host has failed to respond". The next call then succeeds and processing continues...but then fails again later. There seems to be no significance in the number of records processed - this varies from around 19K to 19.1k and fails at different points and with different records. Additionally, when I run the system in multi-threaded mode (Parallel.ForEach() & MaxDegreeOfParallelism == -1), then I'm seeing the same "profile" of failures, but instead of a single record fail followed by the next "working", I'm seeing about 10 failures before it comes back to life - which leads me to believe it's something like the server actually dying and being re-instantiated. Anyone have any ideas? I can batch out the data and get a "new" remoting connection if I create a factory to do so (I'm injecting my remoting client), but this seems like avoiding the problem rather than understanding the cause and fixing it. Ultimately this will be used to process > 1.2M records Cheers guys!
C# has already designed away most of the tedium of C++.
I would pass the whole lot of records at one time. Not only is it going fubar after 19k records, but if you process 1.2M records, imagine how much time you are adding with the network latency between each record.
The difficult we do right away... ...the impossible takes slightly longer.
-
Hi guys, So I have an object that I'm exposing via .NET remoting (for historical / political reasons I can't just drop this and go to something more modern), which I use to process a list of items, one call per item in my list. The remoting server is set up as Single call rather than Singleton, and so my understanding is that I should get a new instance of the server class on each invocation - i.e. for each item in my list. This seems to work just fine for small datasets, but when I push it a little by processing a list containing approx. 200K items, it reaches around 19K processed, I see a timeout, and the call fails with a socket exception - "A connection attempt failed because the connected party did not properly respond after a period of time or established connection failed because connected host has failed to respond". The next call then succeeds and processing continues...but then fails again later. There seems to be no significance in the number of records processed - this varies from around 19K to 19.1k and fails at different points and with different records. Additionally, when I run the system in multi-threaded mode (Parallel.ForEach() & MaxDegreeOfParallelism == -1), then I'm seeing the same "profile" of failures, but instead of a single record fail followed by the next "working", I'm seeing about 10 failures before it comes back to life - which leads me to believe it's something like the server actually dying and being re-instantiated. Anyone have any ideas? I can batch out the data and get a "new" remoting connection if I create a factory to do so (I'm injecting my remoting client), but this seems like avoiding the problem rather than understanding the cause and fixing it. Ultimately this will be used to process > 1.2M records Cheers guys!
C# has already designed away most of the tedium of C++.
