Hello All, I've run into an issue with a TCP memory leak that I'm hoping someone may have seen before. I've scoured the archives but haven't found anything to explain the behavior.
We are running an older kernel (3.2.46) on a high-availability production system that is servicing 10's of thousands of socket connections. What we have seen is that the TCP slab memory allocations are continuing to grow by 10's of MBs per day. Although we have double checked that all of our application socket file descriptors are being open and closed appropriately, there appears to be a continual loss of free kernel memory. We aren't sure yet whether it could be skbuffs that aren't being freed or whether there are orphaned sockets that aren't being closed. We would of course, like to upgrade to a newer kernel to see if the problem goes away, but on this production system it would be very difficult to do. Would anyone know of any memory leak issues in the skbuff or tcp area that may have been fixed since the 3.2 kernel ? Also, would anyone know of any socket accounting issues that may have been fixed within that time frame also ? Thanks for any help you can provide ! -Kevin -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
