v6.28 final RC testing

rextended · Tue Apr 07, 2015 2:57 pm

Please put a disclaimer inside 6.27/6.28:

"ac" models wlan work only with "wireless-fp" package enabled.
using "wireless" package on ac models cause wlan to not be recognized from the system.

RB922UAGS-5HPacD S.N.:5F820473960B

jondavy · Tue Apr 07, 2015 3:36 pm

the key event DEL (delete) stopped working, not delete any item when selected in any table within winbox

DrDeft · Wed Apr 08, 2015 9:14 am

RB1100AHx2 v6.27

http://puu.sh/h5LXn/32a96e0e88.png

There is no cable connected to ether10, but log tells that link on ether10 is up/down often.

MAC adresses on ether10 and LANs_bridge are the same - 00:D4:FF:6D:53:10

No problems are detected, but I can't understand why is it happening with port10 on RB1100AHx2 in log

rpingar · Wed Apr 08, 2015 10:33 am

we get kernle panic on rc20 (and previous) on a p-mp wireless ap using NV2+WDS when there one or more clients that get frquent disconnection.

We opened the ticket Ticket#2015040866000167 where there is the supout attached.

regards
Ros

santa · Wed Apr 08, 2015 11:18 am

Switches mikrotik without igms snooping and DHCP snooping is cheap toys ! Why these toys you produce?

+1

DrDeft · Wed Apr 08, 2015 11:39 am

Switches mikrotik without igms snooping and DHCP snooping is cheap toys ! Why these toys you produce?

+1

bclewl1ns · Wed Apr 08, 2015 11:57 pm

We are still seeing a big problem with Tcp traffic. When NV2 With 6.28RC20 and previous

When using NV2 if any link the the chain is not negotiated at GigE full our upload TCP traffic is 10 Meg. This happens on AC and on N devices. This has created a huge problem with customer traffic.

We tested in the following way.

PTP Testing

PTP A End ---> PTP B End. Bend at 100 Meg A end at Gige 10 Meg up 12 meg down
With both ends and GigE full speed

PTMP AP ->>> CPE . CPE at 100 Meg AP at GigE. 10 meg up 10 Meg down.

If we move to NStream No problem except disconnect. (Disconnects are bad)

sasbibic · Thu Apr 09, 2015 10:14 am

SSTP tunnel fails to connect:

Clinet V6.25 -> Server V6.28rc20 - fails
CLient V6.28rc20 -> Server V6.25 - fails

Client V6.25 -> Server V6.25 - works OK.
Client V6.28rc20 -> Server V6.28rc20 - no test.

I use SSTP connections from clients for management access, so You can imagine where the problem is ?

/Sas

santa · Thu Apr 09, 2015 10:22 am

Why are you testing different using versions of ROS? Just check if "V6.28rc20 -> Server V6.28rc20" is OK.

sasbibic · Thu Apr 09, 2015 10:25 am

Why are you testing different using versions of ROS? Just check if "V6.28rc20 -> Server V6.28rc20" is OK.

How can I upgrade geographicali dispersed routers at once ? If there will be no other solution I'll be forced to do that, but breaking compatiblity betwen versions is very BAD.

santa · Thu Apr 09, 2015 10:28 am

Yes, I agree that this is BAD, unfortunately this is not the first time

. But I know what answer will give you the Support team. Upgrade both ends and then make a test.

sasbibic · Thu Apr 09, 2015 10:42 am

Yes, I agree that this is BAD, unfortunately this is not the first time . But I know what answer will give you the Support team. Upgrade both ends and then make a test.

I'll solve this, but when someone like me relays on management over sstp link and begins upgrade on false end .....

/Sas

computercolin · Fri Apr 10, 2015 10:12 am

CRS125 ingress-port-policer is not working in v6.28rc20. Cannot do more than 1-5Mbps when set.

/interface ethernet switch ingress-port-policer add burst=100k port=ether24-slave-local rate=60M

iperf does 1-5Mbps. Not 60M, 1-5M.
If I delete ingress-port-policer, goes back to normal and iperf does 900Mbps.

paoloaga · Fri Apr 10, 2015 12:49 pm

CRS125 ingress-port-policer is not working in v6.28rc20. Cannot do more than 1-5Mbps when set.
Code: Select all
/interface ethernet switch ingress-port-policer add burst=100k port=ether24-slave-local rate=60M
iperf does 1-5Mbps. Not 60M, 1-5M.
If I delete ingress-port-policer, goes back to normal and iperf does 900Mbps.

This happens to me with 6.24, 6.25 and 6.27 also (not tested on other versions).

docmarius · Fri Apr 10, 2015 1:04 pm

Shouldn't that burst value be higher? The moment it goes in burst mode, it will send at 100k, than back to 60M, again burst mode at 100k and so on. Maybe burst = 100M?

NathanA · Fri Apr 10, 2015 3:00 pm

Why are you testing different using versions of ROS? Just check if "V6.28rc20 -> Server V6.28rc20" is OK.
How can I upgrade geographicali dispersed routers at once ? If there will be no other solution I'll be forced to do that, but breaking compatiblity betwen versions is very BAD.

Who said anything about upgrading actual in-production SSTP servers? Pull a router out of stock and configure it as a temporary/test SSTP server. If you have no spare RouterBoards sitting around, fire up a copy of x86 RouterOS on a spare PC, or heck, even as a VM inside of VMware or something. There are TONS of options open to you that wouldn't require you to install a beta version of the OS on production routers!

-- Nathan

sasbibic · Fri Apr 10, 2015 5:31 pm

I've figured out this by using test equipment, but at some point I'll eventually do an upgrade. I was actuality posting test resoults hopeing to spare someone's problems if upgrading at false end.

/Sas

akschu · Fri Apr 10, 2015 8:02 pm

Fixing the fetch ssl bug would be fantastic since it makes /tool fetch unusable for me:

http://forum.mikrotik.com/viewtopic.php?f=1&t=95576

Got this back from support:

Re: [Ticket#2015040666000483] Bug with /tool fetch and https.
Hello,

Its due to bug in ssl library. Next release will have the fix.

Regards,
Maris B.

Hopefully we will see this in the RC? Normis, can you confirm?

schu

computercolin · Sat Apr 11, 2015 1:35 am

Shouldn't that burst value be higher? The moment it goes in burst mode, it will send at 100k, than back to 60M, again burst mode at 100k and so on. Maybe burst = 100M?

I do not know how burst is working.

What I see suggests 1) CRS interval between rate-limit-check or 2) CRS method of drop packets is not good.

TCP congestion control will adjust and send at maximum limit allowed. All CRS needs to do is 1) monitor ingress rate and 2) drop packets in granular way (RED?).

Instead, it appears CRS opens flood gate wide, closes flood gate completely, opens flood gate again, closes, etc.
This breaks TCP congestion control =o

Also, what I really want to do is ingress-port-policer at 550Mbps and I cannot set burst this high:

jthorpe · Sat Apr 11, 2015 4:39 am

Does anyone know if SSTP will be fixed in 6.28?

I have two CRS125s running 6.27 as SSTP servers and a RB951Ui-2HnD as a client.

The first time I set this up, SSTP worked perfectly, but since rebooting the client, I can almost never get the SSTP to work again. The SSTP connection comes up, but the associated interfaces on the client never enter a "Running" state.

For instance:

----
/interface sstp-client> print
Flags: X - disabled, R - running
0 ;;; SSTP via Internode
name="sstp-out1" max-mtu=1500 max-mru=1500 mrru=disabled connect-to=server1:4430 http-proxy=0.0.0.0:443 certificate=client1 verify-server-certificate=yes
verify-server-address-from-certificate=no user="user" password="password" profile=default keepalive-timeout=60 add-default-route=no dial-on-demand=no authentication=mschap2 pfs=yes

1 ;;; SSTP via Telstra
name="sstp-out2" max-mtu=1500 max-mru=1500 mrru=disabled connect-to=server2:4430 http-proxy=0.0.0.0:443 certificate=client1 verify-server-certificate=yes
verify-server-address-from-certificate=no user="user" password="password" profile=default keepalive-timeout=60 add-default-route=no dial-on-demand=no authentication=mschap2 pfs=yes
----

SSTP sessions are ok:

----
/interface sstp-client> monitor 0
status: connected
uptime: 8m56s
encoding: AES256-CBC
mtu: 1500
local-address: 192.168.255.18
remote-address: 192.168.255.17

/interface sstp-client> monitor 1
status: connected
uptime: 10m47s
encoding: AES256-CBC
mtu: 1500
local-address: 192.168.255.22
remote-address: 192.168.255.21
----

I have tried combinations of pfs=yes and no (making sure they match on the server end as well) to no avail.

I've opened ticket 2015041166000026 with MikroTik regarding the issue.

Is anyone aware of a workaround or if this will be fixed in 6.28?

avenn · Sat Apr 11, 2015 1:38 pm

Hello,

I`ve been following Mikrotik ROS updates for a long time and this one seems to have stalled / slowed down?

Usually when around RC20 released soon.

I hope this means far more testing and taking of time at Mikrotik?

Regards

Aidan

poizzon · Sat Apr 11, 2015 5:20 pm

[admin@MikroTik] > system resource print
uptime: 8m13s
version: 6.28rc18
build-time: Mar/30/2015 15:41:08
free-memory: 40.7MiB
total-memory: 64.0MiB
cpu: MIPS 24Kc V7.4
cpu-count: 1
cpu-frequency: 400MHz
cpu-load: 4%
free-hdd-space: 6.6MiB
total-hdd-space: 16.0MiB
write-sect-since-reboot: 953
write-sect-total: 17952
bad-blocks: 0%
architecture-name: mipsbe
board-name: mAP
platform: MikroTik
[admin@MikroTik] > sys health print

action timed out - try again, if error continues contact MikroTik support and send a supout file (13)
[admin@MikroTik] >

avantwireless · Sat Apr 11, 2015 6:40 pm

Well in the final rc candidate Torch is again broken showing wildly less traffic than is really on the link, to the point that torch is unuseable...

Sending e-mail is still broken for me. The issue is described here:
System Error sending email timeout occurred

Any chance to get that fixed?
This is fixed in version 6.28rc20. Thanks a lot MikroTik!

StubArea51 · Sun Apr 12, 2015 5:58 pm

@MikroTik

Can you give us any timeframe on when 6.28 will be released?

Clauu · Sun Apr 12, 2015 7:38 pm

Let them properly check for problems tickets and so on, better 1major update instead of giving us 2 buggy updates in a month
Btw Happy Easter to all!

avenn · Mon Apr 13, 2015 11:52 am

Clauu

+1000

jdog · Mon Apr 13, 2015 3:52 pm

I agree. I would rather time spent and get a quality release, instead of rushing and releasing garbage.

That said, I would like some more communication as to how the fixes are proceeding.

IE - We now have SSTP fixed, working on BGP this week, and we want to make sure OSPF is good before we release this one.
(Just an example)

StubArea51 · Mon Apr 13, 2015 5:07 pm

I agree. I would rather time spent and get a quality release, instead of rushing and releasing garbage.

That said, I would like some more communication as to how the fixes are proceeding.

IE - We now have SSTP fixed, working on BGP this week, and we want to make sure OSPF is good before we release this one.
(Just an example)

It would definitely be helpful to have some kind of release timeline. I certainly don't want the releases rushed or buggy, but it's very beneficial to know roughly when the new code will come out for testing and maintenance window planning.

bclewl1ns · Tue Apr 14, 2015 4:21 am

I agree. I would rather time spent and get a quality release, instead of rushing and releasing garbage.

That said, I would like some more communication as to how the fixes are proceeding.

IE - We now have SSTP fixed, working on BGP this week, and we want to make sure OSPF is good before we release this one.
(Just an example)
It would definitely be helpful to have some kind of release timeline. I certainly don't want the releases rushed or buggy, but it's very beneficial to know roughly when the new code will come out for testing and maintenance window planning.

+1

avantwireless · Tue Apr 14, 2015 8:24 am

So I have two devices that I just upgraded to 6.28 because of superior ethernet performance... One is an RB911G-HPnD the other is a RB912UAG-HPnD. They are about 1.4 miles apart with signals in the low 50/high 40's. I noticed that they were see a 20 mhz channel negotiation with channel set to 20/40 eC Figured there was interference so I tried Ce. Still couldn't get better than 144mb negotiation... On a whim, I changed one side to eC and left the other at Ce... Now running at 300/400mb/s and a 40mhz channel. What is up with that?

bclewl1ns · Tue Apr 14, 2015 10:02 am

So i was going to upgrade my CCR's to v 6.28rc20 However i noticed that the rc20 is no longer on the filename. Does this mean 6.28 is final and launched??

TomosRider · Tue Apr 14, 2015 10:05 am

Who said anything about upgrading actual in-production SSTP servers? Pull a router out of stock and configure it as a temporary/test SSTP server. If you have no spare RouterBoards sitting around, fire up a copy of x86 RouterOS on a spare PC, or heck, even as a VM inside of VMware or something. There are TONS of options open to you that wouldn't require you to install a beta version of the OS on production routers!

-- Nathan

Well said Nathan. I have similar thoughts on tons of complaints here. Use stable versions that work for you, test the hell out of newer versions of ROS on some backup or spare router...simple as that.

bclewl1ns · Tue Apr 14, 2015 10:07 am

Who said anything about upgrading actual in-production SSTP servers? Pull a router out of stock and configure it as a temporary/test SSTP server. If you have no spare RouterBoards sitting around, fire up a copy of x86 RouterOS on a spare PC, or heck, even as a VM inside of VMware or something. There are TONS of options open to you that wouldn't require you to install a beta version of the OS on production routers!

-- Nathan
Well said Nathan. I have similar thoughts on tons of complaints here. Use stable versions that work for you, test the hell out of newer versions of ROS on some backup or spare router...simple as that.

+1

Tue Apr 14, 2015 2:59 pm

So I have two devices that I just upgraded to 6.28 because of superior ethernet performance... One is an RB911G-HPnD the other is a RB912UAG-HPnD. They are about 1.4 miles apart with signals in the low 50/high 40's. I noticed that they were see a 20 mhz channel negotiation with channel set to 20/40 eC Figured there was interference so I tried Ce. Still couldn't get better than 144mb negotiation... On a whim, I changed one side to eC and left the other at Ce... Now running at 300/400mb/s and a 40mhz channel. What is up with that?

Could you please tell us what center frequencies you were using on both ends? Could you provide support output files to support@mikrotik.com?

avantwireless · Tue Apr 14, 2015 4:35 pm

My bad, I didn't understand that the client follows the AP regardless of client settings on Ce or eC...

On it's way in a few minutes...

Ticket#2015041466000208

So I have two devices that I just upgraded to 6.28 because of superior ethernet performance... One is an RB911G-HPnD the other is a RB912UAG-HPnD. They are about 1.4 miles apart with signals in the low 50/high 40's. I noticed that they were see a 20 mhz channel negotiation with channel set to 20/40 eC Figured there was interference so I tried Ce. Still couldn't get better than 144mb negotiation... On a whim, I changed one side to eC and left the other at Ce... Now running at 300/400mb/s and a 40mhz channel. What is up with that?
Could you please tell us what center frequencies you were using on both ends? Could you provide support output files to support@mikrotik.com?

Bergante · Tue Apr 14, 2015 5:49 pm

Updated a rb2011.

[admin@rb2011-despa] /interface> set 5 l2mtu=1600 mtu=1540
action timed out - try again, if error continues contact MikroTik support and send a supout file (13)
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface> print

(And after the print it seems to be hanging).

I had to change the MTU manually because it lost the original MTU after the update.

And the router has crashed. I can no longer access it.

Latest version grabbed 5 minutes ago.

Bergante · Tue Apr 14, 2015 5:57 pm

Interesting. Lost IP access to it, but it responds to ARP. So not completely crashed.

I'll try to revive it when I can get physical access.

Tue Apr 14, 2015 6:05 pm

Updated a rb2011.

[admin@rb2011-despa] /interface> set 5 l2mtu=1600 mtu=1540
action timed out - try again, if error continues contact MikroTik support and send a supout file (13)
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface> print

(And after the print it seems to be hanging).

I had to change the MTU manually because it lost the original MTU after the update.

And the router has crashed. I can no longer access it.

Latest version grabbed 5 minutes ago.

We just tried the same command and it didn't crash the board. From what RouterOS version you made the upgrade?
What device is connected to that port?
Were you connected to that port you modified?
Are you able to connect to the board via MAC-Winbox connection or Console?

Bergante · Tue Apr 14, 2015 6:29 pm

Update made from 6.27.

The port connects to an OmniTIK running 6.27. I am using MPLS, OSPF and BGP-VPLS. Some time ago I suffered a similar problems, somehow my MTUs got crazy.

Yes, I have been able to do a mac telnet.

A reboot made ospf come back temporarily but something went wrong again. I have problems to print an interface list or even to do an export

[admin@rb2011-despa] /interface> print
Flags: D - dynamic, X - disabled, R - running, S - slave
# NAME TYPE ACTUAL-MTU L2MTU
0 S ether1 ether 1500 1598
1 S ether2 ether 1500 1598
2 ether3 ether 1500 1598
3 ether4 ether 1500 1598
4 S ether5 ether 1500 1598
5 R ether6 ether 1540 1600
6 ether7 ether 1500 1598
7 ether8 ether 1500 1598
8 ether9 ether 1500 1598
9 ether10 ether 1500 1598
10 sfp1 ether 1500 1598
11 S wlan1 wlan 1500 1600
12 R bridge-ektel bridge 1500 1598
13 R bridge-ektin bridge 1500 1598
14 R bridge1 bridge 1500 1598
15 R guest bridge 1500 65535
16 R loopback bridge 1500 65535
17 D vpls5 vpls
18 D vpls6 vpls
19 D vpls7 vpls
20 D vpls8 vpls
21 D vpls9 vpls
22 D vpls10 vpls
23 D vpls11 vpls
24 D vpls12 vpls

[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface> set 15 mtu=1500
interrupted
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface>
[admin@rb2011-despa] /interface> vpls
[admin@rb2011-despa] /interface vpls> bgp-vpls
[admin@rb2011-despa] /interface vpls bgp-vpls> print
Flags: X - disabled, I - inactive
# NAME ROUTE-DISTINGUISHER SITE-ID BRIDGE
[admin@rb2011-despa] /interface vpls bgp-vpls>
[admin@rb2011-despa] /interface vpls bgp-vpls>
[admin@rb2011-despa] /interface vpls bgp-vpls> print
Flags: X - disabled, I - inactive
# NAME ROUTE-DISTINGUISHER SITE-ID BRIDGE
[admin@rb2011-despa] /interface vpls bgp-vpls> export
# apr/14/2015 17:24:45 by RouterOS 6.28
# software id = PCAE-WQ4F
#

Bergante · Tue Apr 14, 2015 6:31 pm

(Sorry I wasn't more explicit, notice the 65535 MTU for the "guest" and "local" bridges which are obviously wrong)

Bergante · Tue Apr 14, 2015 6:55 pm

Whatever, the configuration was somewhat corrupted when doing the update. I think it happened some time ago with another version and I had to rebuild. Alas, as it was something I was doing while paying attention to something else, I didn't end up reporting it. My fault.

I will try to rebuild and pay proper attention when updating the next one.

Bergante · Wed Apr 15, 2015 9:09 am

I managed to go back to 6.27 and it begun working with the same configuration.

The symptom: Unless I stopped OSPF propagating routes (which wouldn't allow BGP, hence BGP-VPLS to work) the router experienced some kind of freeze that made many commmands failt such as a "print" froze.

I have a couple of supout files, I presume you'd like to have a look at them.

Wed Apr 15, 2015 10:34 am

<b>Bergante</b>, please contact MikroTik support with attached support output files, when you have issues with 6.28 version.

Thank you very much for the help.
Please post your replies about 6.28 here.

http://forum.mikrotik.com/viewtopic.php?f=2&t=95883

Who is online