BGP peer goes down, no ACK for 60 minutes

RouterOS General
1

I created a support ticket SUP-135165 on November 21, 2023. It has been left unanswered entirely. I’m posting the contents here without the package attachment for security sake.


We have 3 BGP peers on this CCR1036 router running v6. Two of them are not having any issues. The third peer is a Juniper MX. This third peer started having outages the last 6 months. It looks like the peer will lose connectivity and wait for an hour before it will start working. Doing a disable & enable on the peer does NOT fix it.

The source seems to be a refusal to ACK the BGP establishment from the Juniper MX.

Attached is the packet capture (on the support email) on the CCR1036 that shows the Mikrotik side not responding to the other MX router until packet 2443 (3450.717049 seconds) at which point the peer comes up. The CCR1036 is (look at the email please). I’ve also attached a supout.

We have other Mikrotiks on other archs (ARM, x86) that do not have any issues with the exact same Juniper MX.

Please let me know what we can do to prevent this issue from happening again. It appears to be a bug isolated to this architecture from what I can tell.

2

I’ve seen a couple of Mikrotik employees look at this post. The support ticket remains unanswered.

Does this mean there is no support for the issue? Or the software? Or the hardware? Is this architecture EOL without an official announcement from Mikrotik?

This appears to be the latest info on Tile http://forum.mikrotik.com/t/end-of-sale-and-end-of-life-of-tile-based-ccr/161628/1 (no official Mikrotik posts/information)

3

How do you know they were employees?

4

The fact that this is the only response I get says a lot about Mikrotik.

5

This is a user forum.. Mostly users here.. There are a few employees here but not many..

Have you tried upgrading firmware? Why are you still using v6?

6

It is on the 6.49.10 firmware and RouterOS.

I’m using v6 first because I’m more comfortable with it and second because there are significant issues based on discussion on this forum, Facebook groups, and talking with other users in person.

I spoke with my distributor and they have no stock and are unable to order the CCR1036 in any version. This confirms to me that the architecture is EOL. Based on the complete lack of response from Mikrotik (via email or this forum) I suspect that the software/hardware is being abandoned and will not receive any specific attention/bugfixes to tile.

EDIT: the 1036 is also missing from their website @ https://mikrotik.com/products/group/ethernet-routers

7

It isn’t EoL but it is EoS (end of sales).

The manufacturer discontinued the Tile CPUs, so no more can be made.

This is unfounded. v7 is the updates and bug fixes, you are not installing them. It is well known v6 is getting security updates only.

8

I have 3 support tickets open and have had little response to them either.

9

It isn’t though. I provided my evidence in the post you quoted!

10

Where/what?

They are not being sold/made anymore. They are still being updated.

Latest update was 2023-12-14, latest beta 2023-12-19.

v6.49 is 2+ years old now (2021-10-07), the last update to it, v6.49.11, was 2023-12-11, but they have been minor bug fixes.

Support from any company, especially ones that offer free updates, will have you try it again on the latest version, see if it has already been resolved..

I have a fleet of the Tile CCRs, it is sad that they don’t get all the newest features now, but they are still getting updates, they are end-of-sales, not end-of-life, just like every other Mikrotik board that has been discontinued.

Otherwise, yeah, that is normal for their support..

11

Try rolling back to 6.48 which I had a number of CCR1036 running that and BGP with a number of junipers for a long time.

I have long since migrated all the CCR1036 to 7.12 as it wasn’t that hard.

12

The issue actually started in 6.48 and I updated to 6.49 in hopes it would resolve our issue.

13

The fact that they’re not acknowledging this bug says a lot. ZERO response from anyone at Mikrotik’s official channels. Years ago they got back to me very quickly about an Xbox upnp bug and had it patched in just a few days. Clearly a different experience now…

sys package update check-for-updates
channel: long-term
installed-version: 6.49.10
latest-version: 6.49.10
status: System is already up to date

It is on the “latest version”. According to Mikrotik update server, it is “up to date”.

14

https://mikrotik.com/download

Shows 6.49.11 and 7.13. I didn’t like 7.13 for the changes in fetch, I went back to v7.12.1 for my CCRs.

In the drop down, choose Upgrade (?), to get the latest version.

15

long-term versus stable channel

16

They used to be different versions.. 6.x vs 6.y, now that is no longer the case. Stable and long-term are the same 6.x version now.

There is not going to be any more changes to v6, just minor patches. It was already said that v6.49 is the final v6 version.

17

No-one else seems to be having the issue so you are asking them to fix a “bug” that seems isolated to you.

Backup the config and try 7.1.12 what do you have to lose and it may shed some light?
You would likely have a spare router so even just try it with the spare.

18

How are you able to make that claim? That’s just ridiculous. I have a packet capture proving the bug. I have several Mikrotiks talking to the same Juniper MX that do not have this issue on other architectures but the same version. Please provide evidence for your claim.

This is an active network with active customers. I can not deploy an unreliable version of RouterOS. It sounds like the solution is to do what others are doing: use a different vendor.

19

And how can you make that claim? Have you tested/tried it? What is unreliable about it?

Seriously, you saying there is a bug in the software you are using but you won’t install the software update.

20

Yes, I’ve found bugs in v7 just like everyone else. If you really think v7 is bug free it’s quite obvious this is a waste of time because you’re wearing blinders.

There’s clearly a bug in the software. Again, I have provided my evidence. The versions I’ve tried are ‘stable’ and ‘long-term’ as labeled by Mikrotik.