SMB Performance Issues: Two Unraid Systems "Server" vs "NAS"

[Slarti123]

Hi everyone,
 

I’m experiencing significant differences in SMB performance between my two Unraid systems, which I’ve named "Server" and "NAS". The NAS performs flawlessly, but the Server is slow and frequently experiences issues such as freezing and transfer interruptions.
 

System 1: "Server"

• Motherboard: ASUSTeK PRIME X570-P (BIOS 2606)
• CPU: AMD Ryzen 5 2600X Six-Core @ 3600 MHz
• RAM: 48 GiB DDR4
• Network: Active-backup bonding (bond0), mtu 1500
• Array:
• Parity: 4 TB
• Disk 1: 4 TB XFS
• Disk 2: 3 TB XFS
• Cache Devices:
• Lexar 500GB NVMe (ZFS)
• WDC 1TB SSD (XFS)

 

System 2: "NAS"

• Motherboard: ASUSTeK PRIME A520M-K (BIOS 2604)
• CPU: AMD Ryzen 5 2600X Six-Core @ 3600 MHz
• RAM: 16 GiB DDR4
• Network: Active-backup bonding (bond1), mtu 1500
• Array:
• Parity: 4 TB
• Disk 1: 3 TB XFS
• Disk 2: 3 TB XFS
• Cache Device:
• Intel 512GB NVMe (BTRFS)

 

Network Setup:

Both systems are connected via iperf3 over 10GbE, and network performance appears to be solid.

Transfers via the Web GUI file manager are fast on both systems.

 

However, SMB transfers on Server are incredibly slow, with transfers sometimes freezing or taking forever.

NAS, on the other hand, maintains a stable 110MB/s over SMB with a 2.5GbE NIC connected to NAS’s 10GbE NIC.

 

 

Additional Info:

SMB on Server is the main issue: transfers freeze, drop off, or move very slowly compared to NAS, where everything works smoothly.

I’m not sure where the bottleneck is or why there’s such a difference between the two systems, especially since they’re both similar in terms of hardware and network configuration.

 

SMB Configurations:

 

Server SMB Settings:

Enable SMB: Yes (Workgroup)

SMB Multi-Channel: Yes

Enhanced macOS Interoperability: Yes

Enable NetBIOS: No

Enable WSD: Yes

 

 

NAS SMB Settings:

Enable SMB: Yes (Workgroup)

SMB Multi-Channel: Yes

Enhanced macOS Interoperability: Yes

Enable NetBIOS: No

Enable WSD: Yes

 

Any ideas on what might be causing the issue with Server, or suggestions for optimizing SMB?

I can provide more logs and details if needed. Thanks for any help!

[itimpi]

You are likely to get better informed feedback if you attach your system’s diagnostics zip file to your next post in this thread for the server exhibiting the issue.   It is always a good idea when asking questions to supply your diagnostics so we can see details of your system, how you have things configured, and the current syslog.

[Slarti123]

Good point. Here both of them: "Server" has the issues, "NAS" works like a charm

server-diagnostics-20240920-2309.zip nas-diagnostics-20240920-2309.zip

[itimpi]

The syslog for ‘server’ is flooded with messages of the form:

Sep 20 14:00:01 Server winbindd[8503]:   process_request_send: [nss_winbind (19858)] Handling async request: GETGROUPS
Sep 20 14:00:01 Server winbindd[8503]: [2024/09/20 14:00:01.055003,  3] ../../source3/winbindd/winbindd_getgroups.c:63(winbindd_getgroups_send)
Sep 20 14:00:01 Server winbindd[8503]:   [nss_winbind (19858)] Winbind external command GETGROUPS start.
Sep 20 14:00:01 Server winbindd[8503]:   Searching groups for username 'root'.
Sep 20 14:00:01 Server winbindd[8503]: [2024/09/20 14:00:01.055123,  3] ../../source3/winbindd/winbindd_misc.c:355(winbindd_interface_version)
Sep 20 14:00:01 Server winbindd[8503]:   winbindd_interface_version: [nss_winbind (19859)]: request interface version (version = 32)
Sep 20 14:00:01 Server winbindd[8503]: [2024/09/20 14:00:01.055247,  3] ../../source3/winbindd/winbindd.c:496(process_request_send)
Sep 20 14:00:01 Server winbindd[8503]:   process_request_send: [nss_winbind (19859)] Handling async request: GETGROUPS
Sep 20 14:00:01 Server winbindd[8503]: [2024/09/20 14:00:01.055284,  3] 

Not sure what is causing this, but the ‘NAS’ server is not showing this so tracking down the trigger will probably resolve the issue..

[Slarti123]

Thank you! I had the same observation, but I have no clue how to find the reason for all these winbindd messages. Is there any way to see if it is caused by a plugin/vm or anything else?