ROS 4.4 BUG at x86 with traffic-flow (netflow).

Today I found BUG in ROS 4.4 at x86 with traffic-flow (netflow).
After reboot traffic-flow (netflow) not sent any data to collector (billing system).
It will be stop until:

  1. in terminal run command: ip traffic-flow print
    or
  2. in windox open setting IP->Traffic Flow.
    Very strange, but only after it (after 20 seconds) traffic-flow (netflow) started sent flows.

I chacked it 3 times with different delay from 1 minutes to 20 minutes after reboot, but result the same - netflow start sent data ONLY open setting traffic-flow (from console or from winbox).

Can anybody confirm it?

This might or might not be related: I have a problem with v4.2 where after scheduled reboot connection tracking, SNMP counters and bridge traffic stop working until winbox connects.

it’s your ‘Inactive timeout’, I believe =)

the bug was confirmed (on Jan, 6) by Support, we’re waiting for the fix =)

It’s fixed in todays build

nice )) one more change concerning ‘action=reject’ - and NetFlow will be pretty cool =)

?

it’s in [Ticket#2009121966000044] =)

Currently we have no plans to make it. Maybe some time in the future.

it’s a pity. then NetFlow is still useless for precise data =)

Will ROS 4.5 releasead today?

Problem not with delay 20 sec - I now about ‘Inactive timeout’.
Problem with stoped netflow after reboot until show setting of netflow.
After reboot we lossed data with information client’s traffic for billing system (for me it was 10 hours).

yep, I was told about some Ukrainian ISP that have lost about half a month of NetFlow information %)

and - yes, 4.5 will be released soon, but as always it will be surprise, so nobody knows exact date ))

9 times out of 10, RouterOS is released on a friday evening

haha, as it will be like with 4.4 - then it will be that 1 of 10 %)

:slight_smile:

Bug with netflow (not started after reboot) not fixed in ROS 4.5(x86).
8-(

the same with flow_sequence - no changes… did they forget to include changes to the release? %)

by the way, I have pre-4.6 version, this bug is fixed here at last. unfirtunately, ‘flow_sequence’ still provides useless information :frowning: