M4300 +QNAP +Port Trunking

schumaku wrote:

Once the LAG is correctly working, this is the only are where you can guess or trial. The HASH modes only define on how the frames are distributed over the multiple physical links.

Just want to add that you don't need to match the hash modes used by the switch and the NAS (and you might not see the same options in the QNAP as you see on the switch).

 

The hashing determines the transmit distribution only, so it can be (and often is) not the same in the opposite direction.

 

 

nigelgourley1 wrote:

LAG config, select ports 19/20 on LAG 1 all the others have no ports enabled 
LAg Config Admin Enabled, Link trap Disable, STP enable, if we enable static then the lag is up and we can't connect, if we disable static then LAG is down and we can connect

 

Often misunderstood: On a static LAG, there is no kind of handshake beyond the physical link of each port connection. Where ever you see this "Up", it indicates likely the static lag is configured. The indication does not ensure all the physical links are up (and eg. correctly plugged) - it will always show up as "Up" or enabled.

 

Unless explicitly configured elsewhere, a LAG on a M4300 Managed Switch is by default a static LAG.

 

Slightly different, when configuring a LAG on LACP mode - I would suggest to use that for connecting a NAS or any system supporting LACP (and only use static LAGs for those systems and/or switches which don't)

 

The LACP (Link Aggregation Control Protocol) handshake is the negotiation process where two network devices exchange LACPDU packets (Link Aggregation Control Protocol Data Units) to agree on forming a Link Aggregation Group (LAG) for bundling multiple physical links into one logical link, determining active ports using system and port priorities, ensuring both sides select the same interfaces for load balancing and fault tolerance, a key step for reliable data flow. 

 

This is the only reliable indicator for an LACP link being "up" (or "down").

 

nigelgourley1 wrote:

2 identical ports on QNAP set to Managed switch and 802.3ad and any version/ Balanced RR balances XOR /802.3ad

 

It's not about trial and error.... Both ends of the LAG, the config must match correct - a 802.3ad LACP LAG can't and won't work against a static LAG for example.

 

Balanced RR (Round Robin) or XOR can under some conditions help to adjust the the maximal performance over the LAG in static mode.

 

nigelgourley1 wrote:

Tried the various HASH modes and no joy 

 

Once the LAG is correctly working, this is the only are where you can guess or trial. The HASH modes only define on how the frames are distributed over the multiple physical links.

 

nigelgourley1 wrote:

The QNAP sees it's connected 20G

 

This is an old and outdated idea on how to indicate the LAG, coming from some older Intel drivers supporting multiple links and LAGs. These Intel drivers offered the nowadays obsolete NIC bonding feature, and mostly replaced by the non-hardware dependant Microsoft network teaming drivers. 

 

Note:  There is never a 20G "link", it's a sum of all physical links in a LAG, like 10G+10G. A LAG does never offer the exact same capabilities of a single network link, like e.g. 40 Gbps, 100 Gbps and up to 400 Gbps in enterprise/data center environments 

 

nigelgourley1 wrote:

VLAN membership - all ports untagged, LAG ports all no tagging except 19/20 which are untagged

 

VLAN Sometimes, users doing misconfigurations by accident. The LAG can - especially when running in [T]agged mode - be associated to a wrong VLAN. It's strongly advised to understand the LAG like an additional interface on the switches - so double check.

 

That much said about LAG being "straight forward configurations"....