routes nexthops unreachable, and sometimes it got disconnected without any reason.
I used /ip accounting in the past but switched to netflow which has a lot more capability and can also do IPv6.
I wrote a simple Perl program (using Net::Flow library) to collect the netflow traffic and write it to a tab-separated file where I can run some other scripts to collect some statistic info.
Of course there are nice graphical netflow analyzers available but this was not required in my case (it is mainly for logging in case of complaints).
As I mentioned earlier, Netflow/Traffic Flow also needs a rewrite or hardware acceleration on RouterOS.
Traffic flow consumes all the CPU resources in case of an ddos attack.
Amazing what a fresh kernel, hi wireguard!
Thanks guys
Upgrade from 6.47rc2 to 7beta8 on RB3011 changed Software ID and results in lost license. So no further testing for me.
On-error catches only runtime errors. With syntax errors (which is what unknown keyword, remowed command is) script does not even start.
Not everyone is ‘expecting’ it. Whilst fq_codel / cake would be nice, I don’t feel it should be implmented at the detriment of implementing other features
On a new RB4011 I am having issues running BGP for IPv6. I have a session for IPv4 that works perfectly, but IPv6 does not seem to announce and originate anything.
Here is the situation:
/routing bgp connection
add local.role=ebgp remote.address=x.x.x.x .as=xxxx template=net4
add local.address=2xxx:xxx:xxx::10 .role=ebgp remote.address=2xxx:xxx:xxx::1 .as=xxxxx template=\
net6
add local.role=ebgp remote.address=2xxx:xxxx:1::36 .as=65000 template=net6
Here’s the template
/routing bgp template
add address-families=ip as=xxxxx name=net4 output.filter=""
add address-families=ipv6 apply-changes=auto as=xxxxx instance=default name=net6 output.filter=""
And the static routes
/ip route
add dst-address=xxx.xxx.xxx.x/22 gateway="" type=unreachable
/ipv6 route
add dst-address=2xxx:xxxx::/29 gateway="" type=unreachable
The session gets set up, but no routes are exchanged over IPv6. Again, the same config with IPv4 works perfectly.
Am I missing something, or is this a bug ?
I agree. First priority is to get ROS7 at least as stable as ROS6. And then implement new features.
Personally I wait for SQM like fq_codel, CAKE too. We are trying to provide more comfortable service for our customers. This feature is one that we think improve it (but not 100% sure about it).
Devs, please check if smips version still have problems with ipsec.
On every 7beta I have problems with it. It (I think) maxing CPU to 100% when tunnel is working or looping through some phases of connecting. When this happens I lose connectivity to my router. Only way to restore my access in this situation is to unplug the cable from internet so ipsec will stop connecting. I had similar problem on ROS7b8 (just having problems with connecting winbox to confirm it), my router stopped responding until I have disconnected cable from internet. This is an only problem stopping me to test ros7, because I need an ipsec tunnel everyday.
ipsec config:
/ip ipsec policy group add name=Peak
/ip ipsec profile add dh-group=modp1024 enc-algorithm=aes-128 hash-algorithm=sha256 name=profile_1 nat-traversal=no
/ip ipsec peer add address=123.123.123.123/32 local-address=11.11.11.11 name=peak profile=profile_1
/ip ipsec proposal add auth-algorithms=sha256 enc-algorithms=aes-128-cbc name=peak
/ip ipsec identity add auth-method=digital-signature certificate=local-cert peer=peak policy-template-group=Peak remote-certificate=remote-cert
/ip ipsec policy add dst-address=123.123.123.123/32 peer=peak proposal=peak sa-dst-address=123.123.123.123 sa-src-address=11.11.11.11 src-address=11.11.11.11/32 tunnel=yes
Disabling ipsec peer and policy let me connect router to internet without problems.
On 6.47 I didn’t see that ipsec could eat so much CPU, I could push ~20mbit/s of encrypted traffic between ipsec peers with only 80% of CPU load and it didn’t break any other connectivity.
Thank you for your hard work guys, really appreciate it. I’m happy with every v7 milestone, and this edition is certainly it.
MT should focus on releasing v7 asap after providing functionality similar to v6. Without looking at all the requests for the new features. There will be time for that later.
That is a really pleasant surprise. Never expected to see a 5.x kernel in ROS after TILE architecture has been removed from upstream Linux.
TILE support in RouterOS never depended on the kernel
How about x86 version? Did you checked it?
Even write to usb flash is not working…
5GHz interface on RB4011 wifi model is not working in 7.0beta8. It is in ap bridge mode. Scan shows nothing, devices do not see the SSID. I have 6.47 on another partition on the same device and rebooting to that makes it work again. The 2.4GHz interface works fine in either.
Strangely, my Audience wlan3, which uses the same chip (QCA9984), seems to work fine. I set it up as “ap bridge” and I can connect to it.
My RB4011 cannot even connect as “station bridge” to the Audience, but other devices can, like my iPhone.
When I try connecting the RB4011 wlan1 as either bridge or station bridge to the Audience wlan3 in “ap bridge” mode, it seems to see it at first and tries connecting, but I get “association timeout” in the log.
Whatever the the problem is, it seems specific to the RB4011 wifi model and not the QCA9984 chip. It does not work even if I reset to factory defaults, so it is not a config issue.
Thank you for this - I am having the same issue on my smips device (hAP mini). It is running 7.0beta8. I am not using it as a router so I had ether1 connected but it refused to connect via winbox or terminal (ssh/telnet). I plugged in ether2 instead and rebooted it, and now I can log into winbox and everything works, but terminal still does not (it freezes). The issue does not seem to involve the configuration of the port (on my device ether1 was not set up as an “internet” port in any way), but instead whether there is a cable plugged into ether1 or not.
My experience is that all even numbered WLAN do not work. Check the MAC address.
Add a virtual WLAN and use only the ODD interface numbers
What about x86 version?
Can you solve the issue i wrote before?
Please do not repeat the same thing over and over again!
Tried to load beta8 on a rb1200 and get a kernel panic
RouterBOOT booter 6.47
RouterBoard 1200
CPU frequency: 1000 MHz
Memory size: 512 MiB
NAND size: 64 MiB
Press key within 2 seconds to enter setup..
loading kernel… OK
setting up elf image… OK
jumping to kernel code
Starting…
Oops: Machine check, sig: 7 [#1]
BE PAGE_SIZE=4K RB1200
CPU: 0 PID: 41 Comm: sysinit Tainted: G M 5.6.3-440 #7
NIP: a10758ac LR: a1076f90 CTR: 00000000
REGS: 9ffeff08 TRAP: 0214 Tainted: G M (5.6.3-440)
MSR: 00029000 <CE,EE,ME> CR: 84000208 XER: 00000000
GPR00: a1076f3c 9e491cf8 9f734600 000000f8 00000000 00000000 000000fc 80499990
GPR08: 8047e1e4 a9015114 80480000 9e491cb8 24000208 10028298 a107d000 8038720d
GPR16: a107e094 00000039 803e52dc 00000124 8007297c 00000010 9f5f0a24 00000488
GPR24: a4f8b384 a107b448 9f5f0a24 a1080000 00015000 00000124 a107b284 a107b094
NIP [a10758ac] get_hcfg+0x118/0x124 [flash@0xa1075000]
LR [a1076f90] init_module+0x280/0x408 [flash@0xa1075000]
Call Trace:
[9e491cf8] [a1076f3c] init_module+0x22c/0x408 [flash@0xa1075000] (unreliable)
[9e491d28] [80002a54] do_one_initcall+0x7c/0x154
[9e491da8] [80074258] do_init_module+0x6c/0x1e0
[9e491dd8] [80075dc8] load_module+0x1974/0x1bb4
[9e491ea8] [80076174] sys_finit_module+0x8c/0x94
[9e491f38] [8000b168] ret_from_syscall+0x0/0x38
— interrupt: c01 at 0x77edf204
LR = 0x77eb98c4
Instruction dump:
90010024 48003a79 80010024 80610008 38210020 7c0803a6 4e800020 3d20a108
8129b080 7d291a14 7c0004ac 80690000 <0c030000> 4c00012c 4e800020 3d20a108
—[ end trace 1e39f59b6d518d6a ]—
Kernel panic - not syncing: Fatal exception