The Background: Customer has a disaster network that I’ve been retained to fix.
The Scene: End-of-business, Wednesday last week, Myke is feeling confident.
The Situation: I’ve completed most of the work, and since it’s 5 minutes to closing, I decide to reboot test everything.
The Outcome: Router okay, Wireless Access Points okay, Internet okay, Ethernet Switch… uh… hello? ∗tap∗ ∗tap∗ This thing on?
The Culprit: Linksys SRW224P, a 24 port manageable, VLAN & POE-capable Ethernet switch (and it’s killing me on the inside)

Now I’ve worked on the SRW line switches before, I own a couple SRW2024s and I’ve config’d the 2016s and 2048s for a few clients… but this 224P is going to be the end of me: Define more than 2 VLANs, and most of the configuration is lost after reboot… Yeah: Most. Sometimes it’ll blow away everything, sometimes just the VLANs. Made no sense to me. I’ve even updated the firmware (nice to have a Mac with TFTPd built-in handy) And yes, I’m even going to Maintenance -> Save Config before I reboot – no noticable effect. Called Linksys, all they recommended was trying with MSIE (7 doesn’t work BTW) and updating the firmware. No help.

After hacking at the thing on and off over the long weekend, I finally figured out the problem; the switch has no trouble storing it’s configuration… it’s that it doesn’t like it’s own config-file!

Have some VLANs, click for full view.

!
VLAN database
 VLAN 1 name DefaultVlan media ethernet state active
 VLAN 100 name Private LAN media ethernet state active
                                                ^
% Invalid input detected at '^' marker.
 VLAN 200 name Public LAN media ethernet state active
                                               ^
% Invalid input detected at '^' marker.
!

And it’s not like the VLAN config I’ve concocted is all that wild or difficult…

All I can say to this is WTF?
The dumbest thing about it all is that the switch runs with my desired configuration just fine until I reboot it, so how/why is it storing a bad config? Surely this isn’t actually a unit-specific problem, but yet the support guys claim that nobody else is reporting a problem like this. I imagine some developer at Cisco/Linksys should investigate this, but instead they just want to RMA it. Sigh.

Still like my 2024s tho… Like my 2924s better, but what can you do?