Tonight I remotely upgraded an RB-750. I Remote Desktopped into a remote machine, on that machine I browsed to MT’s download page, downloaded the package, used the browser on the remote machine to WebBox into the router, I uploaded the 5.0rc10 file and rebooted. The RB-750 upgraded to rc10 but I couldn’t reach the machines on the LAN anymore. I WinBox’ed into the remote RB-750 and checked the DHCP leases and sure enough they weren’t bound, they were “offered” for all devices but wouldn’t bind. I WinBox’ed 5.0rc9 into the remote RB-750, in WinBox’s terminal window for the remote RB-750 I issued the downgrade command, and when the remote RB-750 came back up in 5.0rc9 all the DHCP leases were good again.
Tomorrow I’ll try going to rc10 again and see if the problem is repeatable.
On a local RB-750 I’m running rc10 and all is good.
OK, here’s the update. On the remote network I just redownloaded a fresh copy of v5.0rc10 and uploaded to the RB-750 (which was running v5.0rc9) with a different browser just to mix things up, and upgraded and once again DHCP stopped working. All device leases were shown in the DHCP leases window as “offered”. Non of the devices would bind (some TiVo wireless adapters, a PC and a Mac (wireless and Ethernet). So, I again WinBox’ed v5.0rc9 and downgraded and voila all the DHCP devices bound. So this is repeatable.
As I said in my previous post I’m running v5.0rc10 locally with no issues.
Thanks! Just to make sure I understand, of course you mean go back to v5.0rc10, and then get the supout.rif. I’ll have to wait till nobody is home to do that. This router is in use. I’ll try to get to it today. Thanks!
OK, just to be sure, I made the supout.rif under v5.0rc9 which was working fine. Then I upgraded to v5.0rc10 and right away a machine with a wired connection to the RB-750 lost it’s lease but the rest seemed fine. I made a supout.rif under that condition. Then I rebooted the RB-750 and waited a bit and sure enough one by one each device lost it’s lease and didn’t rebind. I made a supout.rif under that condition as well. I think if I had just waited long enough under v5.0rc10 all the clients would have lost their leases without the reboot. I just thought the reboot might speed that process up. I’ll be emailing all the supout.rifs to the address above.
So this problem is highly repeatable. Every time I went to rc10 it happened. Every time I went back to rc9 the problem went away.
A time saver (because this is a remote system) this time was having both rc9 and rc10 already uploaded so after upgrading to rc10 and documenting the problem all I had to do was issue the downgrade command.
According to Sergejs this is a known issue with RC10 when there’s nothing plugged into the switch’s master port:
Hello Greg,
Thank you very much for the report.
The problem is related, that master port is not plugged (when cable is
attached to
the master port, it should work).
We are working on the fix for the problem.