Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi,
I've been hitting this error in my server. I found this log in my Distributionservice > 1 > Log > Cluster_YYYYMMDD.txt
12/19/2012 08:07:18.1179086 Information File system watcher for notification file failed with error: System.IO.InternalBufferOverflowException: Too many changes at once in directory:\\Cnhkgqlv031uat\QlikView\DistributionService.
I have all my configuration files such as config files, log files, distribution server storing at a shared drive \\Cnhkgqlv031uat\QlikView
But my QlikView distribution service will be down once a week. and the error message I found is "System.IO.InternalBufferOverflowException: Too many changes at once in directory:\\Cnhkgqlv031uat\QlikView\DistributionService."
Please advise and comments are welcome.... Thanks~
Hi cklim2011,
It seems I have the same issue; the same log with the same error at shared drive used by clustered distribution service. Everything seems working, but size of log is about 1GB a day.
So any information about this is appreciated
Thanks
Hi all,
Is there any finding on this issue? I also have this error?
Thanks.
Hello,
The issue is defined in bug 47360 and should be solved with v11 SR2, which you may read about in the release notes for this SR as well. If the issue still exists after using v11 SR2 or later, please contact Support.
Regards,
Erik Gustafsson
Hi Erik.
The SR2 didn’t fix this issue, referring
to the internal buffer in .net.
Caused
by memory leak in Microsoft FileSystemWatcher in .NET Framework 4, fixed in
Update 2600217 (http://support.microsoft.com/kb/2600217).
This should be installed with standard Windows Update as part of
"Microsoft .NET Framework 4 Client Profile".
Memory leak most likely is the case. After couple of days working QDS ate all memory without any reason, there were no tasks configured, we had just couple of tests and deleted everything then.
We are going to check .net update from Microsoft.
Memory leak has been resolved with .net4 update. However, logs with messages like these
2/26/2013 13:37:45.2238223 | Information | File system watcher for notification file failed with error: System.IO.InternalBufferOverflowException: Too many changes at once in directory:\\xxx.xxx.xxx.xxx\QlikView\Logs\QPub. |
2/26/2013 13:37:45.2238223 | Information | File system watcher for notification file failed with error: System.IO.InternalBufferOverflowException: Too many changes at once in directory:\\xxx.xxx.xxx.xxx\QlikView\Logs\QPub. |
2/26/2013 13:37:45.2238223 | Information | File system watcher for notification file failed with error: System.IO.InternalBufferOverflowException: Too many changes at once in directory:\\xxx.xxx.xxx.xxx\QlikView\Logs\QPub. |
still exist.
Hi, I also have received support from QlikView Distributor.
We spent like more than 2 months, non of their solutions/recommendations solve the problem. They have told us to do the following:
1) to increase RAM because the publisher is lack of capacity (didn't work)
2) minimize the numbers of QVWs (tested and didn't work)
3) increase the network BIOS limit http://support.microsoft.com/kb/810886 (tested & didn't work)
Eventually I saved myself rather depending on the QlikView support. I removed the sharedDrive, and moved all the files back to Publisher server, and this solved the problem.
Reason why I moved the files to SharedDrive because of the recommended architecture design by QlikView, to enable DR/cluster with ShareDrive. Now I'm back to Square 1 again!
Was this SharedDrive a SAN/NAS?
Patch from Microsoft fixed memory leak, but error message still exists.
We had NAS, which is highly avalable storage system. After back and forth communications with support we were told that configuration with NAS is not supported except that case when "storage must be owned, governed, and shared by a Windows operating system instance". This is from Server Manual document.
We moved files to Windows Share on one of the servers, but error message still exists. Now we are confirmed that is bug, and it will be fixed in further releases of 11.20.