I read that this problem was fixed in a previous firmware, but that may not be the case. I left my home for a short trip yesterday, with my 2312 merrily resyncing, and all seeming to go well. I have a 4-port LAG on my switch for this NAS and the 4-port bond on the 2312 [mode 3-4] which, again, seems to be working just fine. Before leaving, I had copied approx 3.9tb of test files over to the 2312, and set up two shares for test purposes. While at the airport, I started getting hammered with "bond interface offline/online" messages. And I mean hammered, more than twenty since yesterday evening. I haven't changed anything, but am more than willing to believe something has changed in my absence. After all, it did not do this Friday, or Saturday, or Sunday. Unfortunately, I cannot access the 2312 while traveling, because I haven't gotten that far in setting it up. So, I cannot get to my diagnostics & logs. So, I guess I am posting this as an FYI.

schumaku wrote: The 3-4 hash mode is relevant only on the distributed of data leaving the NAS. Yes. The switch would use it's own hash (generally layer-2, and often not configurable). Since that hash mode was configured on the NAS, it must have been set up as LACP - the other bonding modes don't have that setting. The fact that the bond was working when Drunkenspyder was on-site suggests that the switch must also have been set to LACP. That leaves three possibilities eth2 actually has failed (either cabling or one of the ports) the switch rebooted, and the configuration reverted ( as suggested by schumaku ) Some other firmware glitch in the NAS or switch messed up the bond, and the two devices need to be rebooted. When you get back I suggest reconfiguring the hash to use layer 2. Layer 3-4 hashing can overrun a gigabit link, and that normally actually hurts your throughput instead of helping it. There are some special cases where it can give a performance boost, but they likely don't apply. And if they did apply, you'd do better using a static LAG on the switch and round-robin on the NAS.

Forum Discussion

Drunkenspyder

Aspirant

Jan 27, 2020

Solved

Firmware 6.10.2 w/ Repeated “Bond interface bond0 has slave interface eth2 offline.’ Error messages

I read that this problem was fixed in a previous firmware, but that may not be the case. I left my home for a short trip yesterday, with my 2312 merrily resyncing, and all seeming to go well. I have a 4-port LAG on my switch for this NAS and the 4-port bond on the 2312 [mode 3-4] which, again, seems to be working just fine. Before leaving, I had copied approx 3.9tb of test files over to the 2312, and set up two shares for test purposes. While at the airport, I started getting hammered with "bond interface offline/online" messages. And I mean hammered, more than twenty since yesterday evening. I haven't changed anything, but am more than willing to believe something has changed in my absence. After all, it did not do this Friday, or Saturday, or Sunday. Unfortunately, I cannot access the 2312 while traveling, because I haven't gotten that far in setting it up. So, I cannot get to my diagnostics & logs. So, I guess I am posting this as an FYI.

StephenB
Jan 31, 2020
schumaku wrote: The 3-4 hash mode is relevant only on the distributed of data leaving the NAS.

Yes. The switch would use it's own hash (generally layer-2, and often not configurable).

Since that hash mode was configured on the NAS, it must have been set up as LACP - the other bonding modes don't have that setting. The fact that the bond was working when Drunkenspyder was on-site suggests that the switch must also have been set to LACP.

That leaves three possibilities

eth2 actually has failed (either cabling or one of the ports)

the switch rebooted, and the configuration reverted ( as suggested by schumaku )

Some other firmware glitch in the NAS or switch messed up the bond, and the two devices need to be rebooted.

When you get back I suggest reconfiguring the hash to use layer 2. Layer 3-4 hashing can overrun a gigabit link, and that normally actually hurts your throughput instead of helping it. There are some special cases where it can give a performance boost, but they likely don't apply. And if they did apply, you'd do better using a static LAG on the switch and round-robin on the NAS.

3 Replies

Replies have been turned off for this discussion

JohnCM_S
NETGEAR Employee Retired
Jan 30, 2020
Hi Drunkenspyder,

Usually, when you get that error, you should check first the physical connection between the NAS and the switch.

If everything is fine on the physical connection, you may try deleting and then recreating the bond on the NAS.

Regards,
schumaku
Guru - Experienced User
Jan 30, 2020
Drunkenspyder wrote:

I have a 4-port LAG on my switch for this NAS and the 4-port bond on the 2312 [mode 3-4]
What LAG Teaming Mode? What switch? Some other switch brands require using a [Save] to apply the config on a permenent storage - might be that the switch has rebooted, and the LAG can't compe up. What config of the four ports on the switch side?

The 3-4 hash mode is relevant only on the distributed of data leaving the NAS.
- StephenB
  Guru - Experienced User
  Jan 31, 2020
  schumaku wrote: The 3-4 hash mode is relevant only on the distributed of data leaving the NAS.
  
  Yes. The switch would use it's own hash (generally layer-2, and often not configurable).
  
  Since that hash mode was configured on the NAS, it must have been set up as LACP - the other bonding modes don't have that setting. The fact that the bond was working when Drunkenspyder was on-site suggests that the switch must also have been set to LACP.
  
  That leaves three possibilities
  
  eth2 actually has failed (either cabling or one of the ports)
  
  the switch rebooted, and the configuration reverted ( as suggested by schumaku )
  
  Some other firmware glitch in the NAS or switch messed up the bond, and the two devices need to be rebooted.
  
  When you get back I suggest reconfiguring the hash to use layer 2. Layer 3-4 hashing can overrun a gigabit link, and that normally actually hurts your throughput instead of helping it. There are some special cases where it can give a performance boost, but they likely don't apply. And if they did apply, you'd do better using a static LAG on the switch and round-robin on the NAS.