I'm running a Windows 8.1 x64 media server for my living group on my university's campus which serves about 40 people (though not all at the same time). It has a 1000Mbps synchronous connection to all authorized users on our campus. Currently I've set up two user profiles, 'Users' and 'Admins'. Everyone uses the 'Users' logon to access shares, and I use 'Admins' to modify the contents. Active Directory is not enabled. I've set up a multitude of ways to access the server: FTP via IIS, Plex, Website & WebDav via IIS (shows directory listing of our media drive), and shared drives via SMB.
Users login to the webdav directory listing by going to ourwebsite.school.edu/shared%20drive/
Users login to the mapped drive by entering \\ourwebsite.school.edu\shared drive\
It seems that the WebDav and mapped drive access are much slower than the others, sometimes taking upwards of 10-20 seconds to 'resolve' the connection. The mapped drive will 'hang' until completing the connection, but once the connection has been made navigating the drive and transferring files is very quick (~110MBps). I've found that subsequent accesses are also quick, but if I leave the computer idle for a bit the connection hangs upon trying to access the drive.
I found that if I instead connect to the mapped drive with the IP address instead of the UNC or FQDN, the connection resolves nearly instantly, leading me to believe this is a name resolution issue. Though, UNC and FQDN show as local intranet and using the IP shows as Internet (not sure if that matters). I've also had issues with Ubuntu users connecting to the SMB drive using the FQDN, it says 'Name not unique on network'. This is not an issue when using the IP address.
In addition, I RDP into the server while on the same network, and that also takes 5-10 seconds to resolve. The Plex sometimes takes a bit of time to serve the content but showing the media listing is always instant. The website is always accessible instantly.
These are the steps I've taken to try to fix the issue:
- Modified NTLM using GPEDIT
- Modified Registry at HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters -- added DisableTaskOffload w/ value 1 (I rolled this back)
- Turned off Thumbnails (use icons only)
- Disabled Remote Differential Compression (Windows add/remove features) K- Disabled Receive Window Auto-Tuning feature: netsh interface tcp set global rss=disabled netsh interface tcp set global autotuninglevel=disabled netsh int ip set global taskoffload=disabled
- IIS : Note To disable FCNs on the computer, set the following registry entry to the binary value of 0: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\InetInfo\Parameters\DoDirMonitoringForUNC
Usually I'd let something like this go but I use all of these services to populate the drive dozens of times a day, and I feel that something is not optimized or some setting is misconfigured which is causing this to happen. Especially since the connection is 1000mbps. Not really sure where to go from here.
Extra details:
- Using the Windows Firewall
- AV is Bitdefender
- PC has two logical drives, a 128GB SSD for the OS and a 11TB RAID 5 Volume for the media drive
- Server has 16GB RAM, i7-4790K @ 4.0GHz, Intel I218-V Ethernet, GTX 960 GPU
- I've been doing most of my testing from a Windows 8.1 PC but I've been also testing connections with Ubuntu, Windows 10 and Mac OS X VMs
Any help would be extremely appreciated!
