Opened 8 years ago

#8380 new Bug report

Massive Memory and CPU utilization with large dataset Uploads

Reported by: amarcionek Owned by:
Priority: normal Component: FileZilla Client
Keywords: Cc:
Component version: Operating system type: Windows
Operating system version: Windows Server 2008 R2 running under VMware Workstation 8

Description

FileZilla 3.6.0.2

Windows Server 2008 R2 running under VMware Workstation 8
Two Intel CPUs
2 GB memory

Windows 7 Ultimate 64-bit physical machine.
8 Intel CPUs
16 GB memory

I have been trying to use FileZilla to test an FTP server implementation and how it scales using millions of files. My data set contains 10 million 1 KB files in a fairly deep directory structure. What I have found is that after a few minutes, FileZilla becomes non-responsive. A check of the memory sees it climbing very high. Some test runs, I'm able to get it to crash when the memory gets to ~2 GB consumed, probably because its a 32-bit process. Other times, the CPU just spikes to 100% and the process just looks like its hung when looking at the logs on the server side.

This is pretty simple to reproduce on the two machines listed above.

I personally think it has something to do with the queueing mechanism and how much memory is consumed per item. The queue can grow quite faster than files can transfer leading to rather large lists on the queued side.

Attachments (3)

FileZillaMemory3.png (71.6 KB ) - added by amarcionek 8 years ago.
Memory and CPU consumption
FileZillaMemory2.png (71.1 KB ) - added by amarcionek 8 years ago.
Memory and CPU consumption
FileZillaMemory1.png (81.8 KB ) - added by amarcionek 8 years ago.
Shot of the queue

Download all attachments as: .zip

Change History (3)

by amarcionek, 8 years ago

Attachment: FileZillaMemory3.png added

Memory and CPU consumption

by amarcionek, 8 years ago

Attachment: FileZillaMemory2.png added

Memory and CPU consumption

by amarcionek, 8 years ago

Attachment: FileZillaMemory1.png added

Shot of the queue

Note: See TracTickets for help on using tickets.