Memory Leak with GPT drive (3TB)

dan_wilder · Joined: 15 Apr 2008 Posts: 3

I have found that when running a backup to a GPT drive (greater than 2TB) I run out of memory. I have experienced this on three different computers with 4, 6 & 8 GB of ram respectively. I have 2.5 x Ram Pageswap set and yet when processing files (Backup Mirror - Source to Target) to this type of drive it appears to have a memory leak and consumes all physical memory.
- This condition prevents a graceful shut down.
- I have run the profile directly from VVPro, VVEngine with all folders selected.
- I have daisy chained the folders to run in series.
- I have broken them into smaller pieces.
- In each case, the memory gets consumed and not released.
- I do not experience this memory leak condition when the destination is a NAS, or other non-GPT type device.

Any Ideas? Question

Systems (VVPro 2.5-2509/10, with VVEngine 2.1-2102)
1. i7 QuadCore with Hyperthreading (8 CPUs) 8 GB Ram, dual 500GB mirrored raid drives (1-OS & 1-Data), Win7 Ultimate, MS Office 2007, McAfee Total Security.
2. i5 DualCore with Hyperthreading (4 CPUs) 6 GB Ram, 32 GB SSD cache drive, 750GB Main Drive, Win7 Ultimate, MS Office 2010, McAfee Total Security.
3. i5 DualCore (2 CPUs) 4 GB Ram, 500GB drive, Win7 Professional, MS Office 2007, Bitdefender Total Security.

dan_wilder · Joined: 15 Apr 2008 Posts: 3

I think I may have figured this out but will be confident, if the backup process completes.
I had to set unique logs for each individual profile and set it to clear after each run.

My issue may have been the size of the log file giving the appeance of a memory leak. But this begs another question, should the log file be able to consume nearly 100% of the availabe physical memory as it did with my case?

dan_wilder · Joined: 15 Apr 2008 Posts: 3

After setting all logs to 20mb or less, the memory is still being consumed and not released. The memory is not released after a log out and log back in. It is only released after a reboot.

I'm concerned it may be due to file handles and not necessarily to logs. The profiles are working with over one hundred billion files which may result with holding memory threads and not properly releasing them.

Has anyone else had this issue happen? I can reproduce it on several different computers of different brands, configurations all using Pro 2.5 - 2509 and 2510.