What's new in 5.0beta4 (2010-Jun-29 15:48):

*) console - /ping command again returns number of ping responses received,
didn't work in 5.0beta3;
*) added ability to monitor each cpu core usage;
*) added support for SATA,SCSI & USB CD-ROMs in CD installer;
*) fixed problem - IRQ balancer did not use correct CPU core
on non auto MSI irqs after reboot;
*) fixed memory leak;
*) route - new property 'vrf-interface' (for static and dhcp routes),
automatically updates 'routing-mark' of the route to keep it in the same
VRF as the specified interface;

Torrent: http://www.mikrotik.com/download/router ... a4.torrent
HTTP download: http://www.mikrotik.com/download.html

snmp voltage still broken...

*) console - /ping command again returns number of ping responses received,
didn't work in 5.0beta3;

Thank you very much

*) added ability to monitor each cpu core usage;
*) fixed problem - IRQ balancer did not use correct CPU core
on non auto MSI irqs after reboot;

Thanks - boosted my 4core core routers throughput 2.5x, by manually assign interfaces to cores
And now with "/system resources cpu" menu, "/system resources irq" menu makes much more sense.

I also noticed that number of interrupts goes to 0 as soon as interface reaches some certain speed.
It looks like Linux Kernel enable NAPI: http://en.wikipedia.org/wiki/New_API. So on high speed routers i suggest to assign IRQs manually

Hi, Just some feedback

We have a Dell R610 with SAS, but unfortunatly RouterOs doesnt have drivers for this server as yet.
We can work around using flash memory as the hard drive.

Thanks

HI all
works 100% on my new quad core router with 10eth running.
Nice with the cpu and irq meny thanks MikroTik

Upgraded about 30 RB450g/RB433AH/RB493AH and my RB1000 to b4 today. So far, memory leak seems to be plugged.

Opening up an IPv6 route closes my WinBox. This was on a clean install, rb750. I can add/edit routes using terminal just fine.

Memory leak looks fixed so far. I noticed weirdness where an aborted PPTP client session would not delete the IP address and then it couldn't reconnect because the old address was still there.

Making good progress though! I will be very happy to see more services available via IPv6.

Memory leak looks fixed so far. I noticed weirdness where an aborted PPTP client session would not delete the IP address and then it couldn't reconnect because the old address was still there.

Thats been there for a long time with any PPP connection, even on PPPOE from time to time.

I run this script every minute to combat it.

/ip address remove [/ip address find where invalid]

Boot hard drive slow boot about 1 minute and 20 seconds after the card to Newsletter
Also went on the Intel Gigabit NIC do not know whether the drive to improve. .
Before 4.x-5.x has been packet loss network communication failure I can not modify the network card auto-negotiation state or disabled may be able to return to work after the opening of hope to solve

01:58:05 system,info router rebooted
01:59:24 system,info,account user admin logged in via winbox

on x86 wireless wan por,t winbox disconect ... no more then 60 second conection ... on ethernet all ok , that was also an isue on beta 3 ...

all ok on x86 but tourch dont chow port just dst and src address

So far working good on RB450G with User Manager and RB750 in bridge mode. Thanks MikroTik Guys for the better WinBox ! Keep up the good work

no issues here yet on my 433uah, too afraid to try it on my 450's after the last beta...
will see how it goes.

HI all
works 100% on my new quad core router with 10eth running.
Nice with the cpu and irq meny thanks MikroTik

are you using a pc as a router if so
which pci cards are you using in the computer and how many wireless cards are on it
i am trying to work out if 1 pci slot has enough power for 4 R52hn cards
thanks

Got about 50 400 series units upgraded and running SSTP to my RB1000 on 5.0b4. So far so good. For my purposes at least.

HI all
works 100% on my new quad core router with 10eth running.
Nice with the cpu and irq meny thanks MikroTik

are you using a pc as a router if so
which pci cards are you using in the computer and how many wireless cards are on it
i am trying to work out if 1 pci slot has enough power for 4 R52hn cards
thanks

I'm using a x86 berabone 1U router HW that are having 10 ethernet direct on the motherboard.

Has anyone tested N in 10mhz and 5mhz channel sizes? I am interested in deploying it in a PTMP AP scenario.

110 SSTP connections to a RB1000 on 5.0b4 w/ client/server certificate verification, 5 connections local, CHAP authentication, 105 connections RADIUS CHAP authentication, 2048 bit certs w/ the following remote HW:

2x RB433
9x RB433AH
91x RB450G
8x RB493AH

30ish running hotspot, 106 running w/ non-caching proxy.

So far, so good. Disabled OpenVPN now. Hoping to NOT have random crashes on the 1000 that I was having before.

Still looking to find out what level/cipher of encryption SSTP uses. Anyone?

Then I Try to make ssh from router (5B4) - I see next "debug" :

client version: SSH-2.0-OpenSSH_4.3p2 Debian-9etch2
transport state: 0 --> 1
kex algo: diffie-hellman-group-exchange-sha1,diffie-hellman-group14-sha1,diffie-he
llman-group1-sha1
host key algo: ssh-rsa,ssh-dss
enc algo CS: aes128-cbc,3des-cbc,blowfish-cbc,cast128-cbc,arcfour128,arcfour256,ar
cfour,aes192-cbc,aes256-cbc,rijndael-cbc@lysator.liu.se,aes128-ctr,aes192-ctr,aes2
56-ctr
enc algo SC: aes128-cbc,3des-cbc,blowfish-cbc,cast128-cbc,arcfour128,arcfour256,ar
cfour,aes192-cbc,aes256-cbc,rijndael-cbc@lysator.liu.se,aes128-ctr,aes192-ctr,aes2
56-ctr
mac algo CS: hmac-md5,hmac-sha1,hmac-ripemd160,hmac-ripemd160@openssh.com,hmac-sha
1-96,hmac-md5-96
mac algo SC: hmac-md5,hmac-sha1,hmac-ripemd160,hmac-ripemd160@openssh.com,hmac-sha
1-96,hmac-md5-96
comp algo CS: none,zlib@openssh.com
comp algo SC: none,zlib@openssh.com
lang CS:
lang SC:
packet follows: 0
agreed on: diffie-hellman-group-exchange-sha1
agreed on: aes192-cbc
agreed on: aes192-cbc
agreed on: hmac-sha1
agreed on: hmac-sha1
agreed on: none
agreed on: none
died with signal 11 on Tue Jul 6 09:29:08 2010

With ROS 4.10 - all OK.

Any Ideas?

Had an autosupout this morning. All SSTP connections were reset. CPU is pegged at 100% on my RB1000 after the crash and until reboot.. Memory usage is quite a bit higher than I remember it. Maybe memory leak isnt totally fixed yet? Either that or SSTP takes substantially (~50-60mb) more RAM to operate?

What The Function!!

We started to deploy VoIP appliances on our network, with an Asterisk Server and voip phones / ATAs at client side..
Everything works fine (Thanks nstreme).
We needed to make Fax over IP work too..
Beta tests made in our office after some head bashing were successfull.
We found a Customer tester that needed a Fax..
We took the ATA and the Fax and went to the customer and it didn't work...
Client had RB411 ROS3.30 with masquerade.
Back to the office.. it works again..
We reproduced the enviorment of the client and didn't work (With a RB112 )
The only difference was the ROS version...
The office had 5.0beta3 (for testing)
We put 5.beta4 with the standard CPE configuration to the RB112 (that was working smoother than 3.x and 4.x) and everything works..
I didn't see any change at changelog though..

Upgrading to ROS 5.0beta4 we fixed problems with masquerade and Faxing over IP
It should be mentioned...

Upgrading to ROS 5.0beta4 we fixed problems with masquerade and Faxing over IP
It should be mentioned...

[/quote]

Improved connection tracking or sip helper is not in the change log either....

Improved connection tracking or sip helper is not in the change log either....

Changelog says that the kernel version was upgraded. Probably has more to do with that specifically.

On-topic, had another RB1000/SSTP lockup today. It just dropped all SSTP connections and none of the remotes could reconnect until i restarted the 1000.

I've run into a problem with the SSTP server in the new v5.

The SSTP server runs on port 443 and it has to run on this port as Windows clients can only connect on port 443 as there is no configuration option in Windows to connect to any other port as this is considered the default for HTTP SSL.

The SSTP server binds itself to every IP on the router on this port so there is no way to use port 443 for anything else.

There are many people, I'm sure who run a web server requiring SSL on their internal network and DSTNAT external IP:443 to their internal servers.

So now I am stuck.

If I want to use SSTP for Windows clients I must use port 443 on the SSTP server settings page.
If I want to run my web server using SSL I must forward port 443 from my router to my internal LAN

So it's one or the other, either I can run SSTP or I can have an SSL web server. Really SSTP needs to bind itself to a specific IP address as SSL is not specifically required for just that purpose, there are many other systems enforced to run on port 443 that this stops working.

Is it possible to bind the SSTP server to a specific IP address.

The problem is that it has ro run on port 443 as this is the only acceptable port that a windows client can connect into as there is no way to change this in windows.

It is also the port that HTTP SSL must run on for secure web sites.

What this means is that as soon as you enable to SSTP server on port 443 the DSTNAT rules for my SSL web server on my LAN stop working.

It basically means you can either use SSTP or use SSL web servers, you cannot use both as they both HAVE to be tied down to port 443.

If we were able to bind SSTP to a single IP address then we would not have this problem.

Thanks, Dom.

Windows server configured as sstp server has way how to determine if request is HTTP over SSL or SSTP. So it's possible to have SSTP server and webserver with SSL on same IP and port. More on RRAS blog in part where sra_{BA195980-CD49-458b-9E23-C84EE0ADCD75}/ is mentioned.

Found a workaround for the Mikrotik ... with thanks to <Chaos> for his input.

Run the SSTP server on port 4430 (not 443) which will bind it to both IP addresses on the router 1.1.1.1 and 1.1.1.2.

Use dstnat to forward traffic for 1.1.1.1:443 to the internal web server at 192.168.1.1:443
Use dstnat to forward traffic for 1.1.1.2:443 to the SSTP server at either 1.1.1.1:4430

A bit ugly, I admit but this does seem to work once the relevant firewall rules all added. Still would be nice to bind SSTP to one or more IP addresses but hey this is a good start.

Cheers, Dominic.

I keep asking Mikrotik to allow binding services to ports. NTP, SNMP, PPP tunnels, they all need a way to tell them not to listen on every single IP address the router has. It fixes two things, it allows you to save resources (not listening on *.*.*.*) and it also allows the service to reply on the same IP address no matter which interface it came in on. UDP replies are especially hard because they don't seem to be able to be mangled (not part of the initial connection it seems?)

Working great, no more RX drops with the new irq balancer!

The SSTP server binds itself to every IP on the router on this port so there is no way to use port 443 for anything else.

You can just block the port in the firewall 'input' chain, or really just allow 443 to only the desired IP and then block everything else. I tested this and it does work, I was able to connect to the SSTP server and have HTTPS dstnat on another IP.

Upgraded v5 beta2 to beta4 on my rb1000, some defects found:

- Winbox doesn't remind settings any more. Every time after close the winbox and open again have to set my specific settings again. (Columns, column size)

- Winbox tool: Traceroute. After finishing traceroute to IP address the winbox doesn't show any more the end destination is reached. Status bar stays empty. Does user has no clue if test is still running of finished if the end IP address is unknown (url input)


- Request for winbox upgrade: Ability to open new winbox in pull down menu option after selecting registered client or neighbour. It works for telnet session, why not some code embedded that opens a new winbox session with the selected IP? Would save winbox users the need to copy and save the IP address of a registered client fist, manually open a new winbox session and paste the IP to open a new session.
This request made before and supported by several forum users. Please!

rgds

About every two days my SIP VoIP cannot register with the server. It's probably related to the possible change in the connection tracking as it's been working fine before.

Asterisk shows the registration request has been sent but it never completes.

A reboot of the router solves the problem instantly. Before I have had the SIP service port enabled - just now I disabled it, we will see if it has any effect.

WirelessRudy

1. Can't repeat, my winbox remembers them correctly no matter how I close it
2. Yes, can repeat, will fix.
3. Winbox.exe should be opened, that could bring problems as Winbox the program, and Winbox the loader are different things. There is no predefined Winbox.exe location.

WirelessRudy

1. Can't repeat, my winbox remembers them correctly no matter how I close it
2. Yes, can repeat, will fix.
3. Winbox.exe should be opened, that could bring problems as Winbox the program, and Winbox the loader are different things. There is no predefined Winbox.exe location.

OK,
1. True, it does remember last setting of which windows were open. But, in older version, even if a window disappeared because other session (other PC) had other windows open and was closed last, when winbox was opened in first PC again and sub menu windows were opened again most of the columns were the same as when that PC sessions had is set before.
Now they are just back to default setting.

2. thanks

3. Is there no way user then can import a default short-cut into an existing winbox session that opens loader automatically after inititating such by some actions in running program?


Like many other users that not (yet) use "Dude" program winbox is the main tool to work and monitor the network. Many small users run a PC with a winbox session open with several certain sub-windows opened.
Like I do, I run the WAN and LAN interface traffic graph together with a netwatch tool and the simple queues to see who is making most use of the network etc.
These settings are always the same on that PC and on a big monitor just visible for visual check.
At the same time on another PC do other controls things. Configuring, fine tuning, managing etc. Here other submenus for same router is used and if same subwindow in use then with different columns visible etc.

Now every time when you close a winbox session, or when the PC's are shut down (power cut, just every night before going to bed etc.) only the last running winbox session with its settings is recorded.
Thus next time (day) when PC's are started up again/winbox is opened again, the settings of the last closed winbox is used.

So my daily routine is to make my monitor PC showing the right winbox format (sub windows, column and column size etc) again before I can see what is happening with my network and I can start the day. Not a big deal but on a daily base becomes a bit irritating...

Winbox does have a pre defined setting stored in its code somewhere (since each meny/sub-menu's window opens in a certain default format) it would be nice if users were able to set their own set of pre defined winbox settings.

Now please don't mention the option of using the "dude" program. I tried it a couple of times and ran in too many problems so left it. And I do happen to know several other managers of MT based networks never came into using it.

It would be very nice if winbox program could be developed more into a even more user friendly program.

yeah! how does The Dude open new WinBox session? =) there's no need to have WinBox Loader on local computer

It has it built-in

so what's problem to build Winbox Loader into WinBox? =)

Size matters.

Size matters.

to whom?

<40kb is almost nothing. Nowadays with PC's running MB big programs this is hardly an argument....

It sits inside the npk and inside the RouterBOARD. A bunch of DLLs. Size does matter.

Maybe its time for a second generation RouterOS running on more space, not compatible with the older routers. A separate branch containing more functionality, because now memory is cheaper.

It sits inside the npk and inside the RouterBOARD. A bunch of DLLs. Size does matter.

it IS already in RouterOS - you can download it via web-interface

considered that a complete npk package is several Mb's or even some of the single packages are MB's big I don't think a couple of kb's make a difference any more.

A simple code that give user the ability to couple a short cut to its winbox.exe program with a given IP (or mac)taken from running session in a pull down menu for opening a new session shouldn't need so much more code that the program becomes too big?

And what about the other changes we would like to see and that have been mentioned before.
Why a minimum size of menu and sub-menu windows.? Sometimes I want windows even smaller or I have to overlap windows in my main window to get all info in one main router window.
Why default column listing in several sub-menu windows?

Example:
If a winbox -Wireless / Registration - session is opened I get a default setting. This each time when the router has been rebooted. Who decides which are the wanted options here?
I have no interest in "Last activity" or "Uptime" or if the registered device is an "AP" or "WDS". But these are the default settings I cannot change.
What I want to see for example is "Ack. Timeout", "Tx Signal Strength", Tx/Rx CCQ amongst others so I can see at one glare how the radio links are performing.

There are plenty other examples where the pre-set default settings have to be changed each time again (many times when I log into a client CPE I have to set the windows settings again. Apart from my upgrades and thus reboots many users switch their routers off over night and thus are settings back to default.) and that becomes a bit frustrating at times...

Winbox is a very handy and smart tool but I am sure that without too much effort some programmer of MT can make it a even more useful tool!
Lets not forget that I still prefer MT over other make because one of the reasons is the winbox tool that gives you a GUI in windows but that doesn't rely on htm.
One of the weak points of MT's big competitor UBNT is that all their units basically have to be access with a browser session, which is nice, but in real network management can be a pain.
(Yeah I know, they have telnet too, but I could never find the proper commands for them...)

So please MT, spend some of your resources on the winbox tool to make it even better then it already is!

OK, we will think about it. Thanks for the suggestions.

OK, we will think about it. Thanks for the suggestions.

Ok, we'll be waiting with anxiety!

Has anyone noticed that the ping command does not return any response when used in v.5.0 beta 4 under some conditions.

I have a script ... :log info ("ping result=" . "$[/ping 8.8.8.8 count=5]") which should put "ping result=5" into the log file if it runs ok.

If I run this script using "/system script run 0" - it returns "ping result=5"
If I run the script using the "Run Script" button in Winbox it returns "ping result="
If I run the script using the scheduler it also returns "ping result="

It looks like under the "Run Script" or Mikrotik Scheduler that the /ping command does not return any results. It's sending my failover script into meltdown as everything is considered to be failed as no pings are returned. I don't have a v4 router to try and I only started beta testing at v4. It would be handy if someone could test on a v4 platform for me.

Thanks, Dom.

That last sentence had too many v4's in it and should read ...

I don't have a v4.xx router to try and I only started beta testing v5.0 at beta4. It would be handy if someone could test on a v4.xx platform for me.

ok, Mikrotik have confirmed this is a known problem to be resolved in v5.0 beta 5. No timescale at the moment for the beta 5 release.

OK, we will think about it. Thanks for the suggestions.

[user@xxx] > export
.......
/system resource irq
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
set (unknown) cpu=auto
....

Does this correct?