Question : Error #3035 System Resurce Exceeded

We have a purchased application we have been running for several years.  The application was designed to function in multi user mode.  It is licensed per user and we currently have a maximum of six users It was developed in MS Access and we have MS Access Runtime loaded on a Terminal Server to run this application.  We have another application which uses SQL 2005 plus our users are using MS Excel and MS Word on the Terminal Server.

For several months users have occasionally been receiving error messages that say "Error #3035.  System Resource Exceeded".  After many changes we are still receiving these messages and are hoping for help to solve our problem.  Here is a recap of some of the steps we have taken.

Our Terminal Server was running Windows Server 2003 32 bit with 4 Gb of memory and a RAID1 72 Gb drive formatted with Drive C: and Drive D:.  We were frequently running our application and storage drive down below 10 Gb free space.  We rebuilt the server with Windows Server 2003 64 bit and upgraded the memory to 12 Gb.  We converted the Drive system to RAID5 and now have over 40 Gb free in each of the drives C: and D:.  On the old server we were running Symantec Endpoint Protection.  We saw numerous articles suggesting that this could be causing some of our problems.  We reloaded Symantec on the rebuilt server while we looked for an alternative and finally chose Panda Security for Business. We unloaded Symantec Endpoint Protection and ran Symantec Clean Wipe utility.   Within Panda we have set parameters to exclude the following folders, all located on the D: drive, from being scanned: APPDATA,  D:\APPDATA, APPLICATION, D:\DOCUMENTS AND SETTINGS\%USER%\APPLICATION.  

The vendor's customer support recommended running a registry cleaner.  I downloaded and ran CCleaner.exe after research to find a cleaner that would not cause additional problems.  I am now scanning twice per week and finding minimal items to be cleaned.  

Another item I found mentioned MaxLocksPerFile in the registry.  I had long since changed the value from the default of 9500 to 100,000.   I have since increased it to 1,000,000 with no change in performance.  I have done that under the key HKLM\SOFTWARE\Wow6432Node\Microsoft\Jet\4.0\engines\Jet3.x and  HKLM\SOFTWARE\Wow6432Node\Microsoft\Jet\4.0\engines\Jet4.0     I just looked at an article at http://www.velocityreviews.com/forums/t489623-dao-3035-error-update-problem-under-windowx-xp-x64-system-resource-exceeded.html where I see a suggested setting for Threads at these locations with a value of 6 whereas my system has a value of 3.  I will not be changing that without confirmation that it will improve my performance.  

As mentioned before this is an application we purchased from a vendor.  The vendor has several hundred customers who do not have this type of problem.  Several of them have many more simultaneous users than we have and many have larger databases than we have.  Our database is approximately 1.7 Gb in size right now.  Almost everything in this application is done with batches.  Most of our errors come while posting cash receipts from the batch.  The batches have a field for STATUS and 3 of the statuses are "PENDING", "PROCESSING", and "COMPLETE"   When we start to post a batch we change the status from "PENDING" to "PROCESSING" and when the posting is complete it should change to "COMPLETE".  Sometimes even though we have the Error 3035 during the posting we find that everything did post properly and the status is properly changed to "COMPLETE".  But sometimes that status does not change.  When that happens we must check to see that the receipt properly posted both to the General Ledger and to the Customer's A/R history.  So far we are usually finding that it had posted properly and we just have to change the Status Flag.  In order to access the table through Access Run Time we have to call Vendor Support and ask for a password of the day and work with the support person to be sure we properly change the flag and delete the transaction detail records.  Just a few minutes ago a different user found a batch for handwritten checks that posted through to all subsidiary journals but still had a flag that it was "PROCESSING" and it did that without producing any Error message.  I will also comment that I have looked at the Event Viewer and have found no events at the time of the problem in the Application area.

We have been struggling with this problem for a long time and any help would be greatly appreciated.  

Answer : Error #3035 System Resurce Exceeded

Hmm, it seems you are on the right track. Such hardware problems are not easy to fix or even trace... I remember the time (15 years ago) when SCSI controller did not write some single bits properly to database when the disk traffic was too high. It was SQL Server already but we probably did not use data parity check on HDD.