The Graveyard - Products No Longer Supported > DGS-1224T

Does the DGS-1224T reboot when its configuration changes

(1/1)

GeneNZ:
Hi there,

I'm currently planning a project which will require us to enable Jumbo Frames on our DGS-1224T (Rev D).  As the switch is currently in a production system, I would like to know if I change the configuration from the web interface, would it require a reboot of the switch to take effect?

More specifically, I'm wondering if enabling Jumbo Frames on the switch will cause the switch to reboot itself to take effect, or will the effect occur immediately (much like an on/off switch).

Similarly, would changing the management IP address/subnet mask, require the switch to reboot to take effect?

Thanks in advance.

Gene

Fatman:
Jumbo frames is a rather low level change, expect a reboot.  Even if the switch didn't require it I would.

GeneNZ:

--- Quote from: Fatman on August 17, 2009, 08:34:40 AM ---Jumbo frames is a rather low level change, expect a reboot.  Even if the switch didn't require it I would.

--- End quote ---

Thanks for your response.  I'm wondering if you also might be able to help me solve my understanding of Jumbo Frames.  At present we have a couple of DGS-1224T as above (both Rev D Hardware).  On both those switches, the switches (based on the webinterfaces) have their jumbo frames disabled.  The end devices are both using NICs that support jumbo frames, but the jumbo frames are disabled by default on both. 

If I do a ping from Windows (using "ping 192.168.101.105 -l 9000") or from linux for that matter, I would expect that frames to be dropped, since the switch doesn't support Jumbo frames as yet.  Similarly, I would expect that both end device NICs will drop the frames since again they haven't been enabled for Jumbo Frames.  But instead the pings work fine.  The reply is:

Reply from 192.168.101.105: bytes=9000 time<1ms TTL=64
Reply from 192.168.101.105: bytes=9000 time<1ms TTL=64
Reply from 192.168.101.105: bytes=9000 time<1ms TTL=64
Reply from 192.168.101.105: bytes=9000 time<1ms TTL=64

Understandably this is confusing me.  My only reasoning for this is that the end NIC's are segmenting the ping packets into 1500byte blocks, and recombining them at the other end NIC at 9000 bytes.  Would I be right in thinking this?

Thanks in advance.

Gene

GeneNZ:
Just answering both my questions:

1) It appears that just enabling Jumbo frames from the web interface does just act like an on/off switch.  I've tested this by setting up two end-to-end NICs (that support Jumbo Frames).  I performed a standard ping with Jumbo Frames turned off on the NIC and switch, and it worked.  Enabled Jumbo Frames on the NIC's, but kept the Jumbo Frames turned off on the switch, and a ping sized at 9000 bytes failed to return a reply.  Then enabled Jumbo Frames on the switch and the same 9000 byte ping was successful.  The switch was not power cycled in anyway.  I may still reboot the switches to be safe. 

2) Second question I asked was why a ping of 9000bytes does indeed work when the NIC/Switch doesn't have Jumbo Frames enabled.  This was very much a basic question since I'm new to Jumbo Frames.  Turns out each 9000 byte frame is fragmented into 1514byte fragments and sent to the destination.  When you enable Jumbo frames, a single ~9000 byte frame is sent.  I checked this using wireshark.

Thanks again.

Navigation

[0] Message Index

Go to full version