Hi, for test reasons updated a rb133c as wirless client with 3.0rc9.
The AP was still running 2.9.46.
The rb333c worked normally before with 2.9.46 and even with 3.0rc9 without any config changes.
Did do a remote reboot (winbox) of that client rb133c but hardly anything happened.
Only after more then half a minute IP traffic stopt and I couldn´t even reach the unit anymore.
Not from within the AP (mac telnet) not by winbox. Even a ping from the AP to the CPE didn´t come back anymore.
But still had contact with the client through Skype while he was still running a download. The registration table in the AP still showed the CPE with signal strengths and traffic over it.
Only after some minute or so all traffiic died, including Skype connectivety. But the AP still showed the unit in the register.
After some other 10 or 15 secs the unit finally dropped from the register table and according client connectivity was lost on his end too. So, finally the unit rebooted…
His Skype came back up within 5 secs, as well as the unit in the reg. table of the AP.
But is still took another 30-40 secs before IP traffic was restored in full and I could Winbox in or ping the unit.
The whole rebooting process took about 4-5mins!
Before when the CPE still had 2.9.46 the whole remote rebooting process took no more then 20-30 secs in total..
Was this because heavy traffic was on the wireless link or is it a (minor) problem yet to be solved?
Anybody experianced same problem? Mayby something to look at by MT?
the rb133c runs in a routed wirless network without encryption and has dhcp server on its ether1 interface.
So the running packages are:
advanced-tools
dhcp
hotspot
ppp
routerboard
routing
securtiy
system
wireless
OK, I can disable ¨hotspot¨ but the rest is needed for the rb133cd to run smoothly I presume?
I´m only not sure on the ¨security¨ package. Does it cover all security; i.e. encryption for the wireless, passwords and user accounts, vpn etc? Or only the wireless part?? In that last case I can disable that too?
But apart from the hotspot, these are the packages running on all my CPE’s and they (all running 3.9.x) haven´t got the slow reboot issue.
Are you using routing protocols(OSPF,RIP) on this board? If no then disable the routing package.
You can also disable the advanced-tools package if you are not using those tools.
You can also disable the security package - it is only used for winbox secure connection and ssh.
You can also disable the hotspot package.
If you don’t use dhcp, disable that also.
OK, that´s actually the first time somebody explains to me a bit more clearly which packages to use when and which ones not.
I agree, the function of each package can be distilled by studiing the ROS ref. manual, the Wiki and forum. But actually somewhere in the manual it could be explained in a separate chapter (¨ROS summary; a brief prescription of MT ROS menu commands and winbox commands¨). I think a lot of new MT users would highly appreciate such an summary.
Anyway,
If I´m only using routed network (routing table information) but not RIP and OSPF so I don´t need the routing package? Never new this.
Advanced Tools? What are these tools? Are these the tools used under the ¨tool¨ command? In that case nobody should remove these? Who doens´t want to do a occasional ´ping´, ´traceroute´ ´Torch´, ¨Netwatch¨ or whatever? So I do not clearly understand why you think the ¨advanced-tools¨ package can be disabled? Explain which tools it has embedded?
Security package? Never understood it’s only for secure connectivity by Telnet or winbox! I´ll bet lots of users would think it also covers web encryption, VPN and other usage of the OS!
The rest is clear.
We’ve seen a pretty dramatic difference in how rc6 - rc9 responds to commands in Winbox vs 2.9.x.
rc9 is better than rc6, but still pretty “clunky”. Takes a long time to log in, a longer time to polulate information, and a very long time to actually reboot after issuing the reboot command.
We have seen this exact behaviour on a number of different units. They all pass data just fine however.
As an example, we have been able to log into an RB133 ten minutes after issuing a reboot command and still are able to browse settings etc, albeit very slowly.
It feels like management threads in 3.0 Winbox are operating at a lower priority than they do in 2.9.x and they have to wait for other things before being allowed to execute.
This behaviour is completely reproducible in our environment. We compared the default install that loads all packages vs a “stripped-down” install with very few packages and see no appreciable difference.
Very strange. Feels like swimming in treacle. But passing data through the RB is not affected at all.
In addition, the longer the unit has been running, the more likely it will exhibit this sluggish management behavior. Also, it can take two reboots (rather than just one) to clean the unit up so management feels more like 2.9.x.
We definitely see CPE more likely to be affected than APs. Example, 133c station is more likely to get sluggish than a 133 AP.