Memory leak on Windows Server 2012 R2
Has anyone come across a memory leak when running the TeamCity server on Windows Server 2012 R2 ?
What we're seeing is that the Windows FilterManager (FMfn) driver is gradually allocating more and more nonpaged pool until the system slows and finally falls over (when the npp size is ~95% of physical memory). The jumps in pool allocation are sporadic across the day with a *slight* correlation with "heavy" TC actions such as copying project sub-trees.
The software running on the server is TeamCity 2017.1.2 using 32-bit Java 1.8 (with -Xmx1024m), mercurial clients, plus a mercurial repo for the versioned settings feature. There is no local agent and the database server & artifacts storage are external.
At the moment we're just looking to identify suspects...
Please sign in to leave a comment.
Hi,
it's hard to point out a culprit only from the set up. We aren't aware of any related to yours in particular, so in order to investigate further, we'd suggest taking a memory dump. Then you probably will want to send the info to us using the "Submit a request" button above. Follow the steps here for both taking the dump and uploading a large file:
https://confluence.jetbrains.com/display/TCD10/Reporting+Issues#ReportingIssues-OutOfMemoryProblems
https://confluence.jetbrains.com/display/TCD10/Reporting+Issues#ReportingIssues-UploadingLargeDataArchives
Ok, thanks.
We have now identified the problem; in case it is of help to anyone else we had to install this Windows hotfix
https://support.microsoft.com/en-us/help/3051690/system-becomes-unresponsive-when-file-system-minifilter-drivers-are-in