Exactly this below.

We used 1072 units as core Edge/BGP and OSPF/MPLS only, no connection tracking.

We upgraded from 1072 to 2116 units v7 on all edge units and 2004 v7 on all 
core units and also all access units using connection tracking. The 2004 units 
are great little processors of traffic, when they are in stock.

From: AF <af-boun...@af.afmug.com> On Behalf Of Trey Scarborough
Sent: Wednesday, April 12, 2023 9:06 AM
To: af@af.afmug.com
Subject: Re: [AFMUG] Tik 1072 watchdog reboot bug


Its a known hardware issue with connection tracking enabled and hardware 
offload. It has a hard limit to the number of connections it supports that is 
pretty low. Its high enough you won't notice till you get significant traffic, 
but low enough it is a common issue. The fix is to turn off connection tracking 
I know this isn't the best solution, but its the only one that works. This and 
the hardware availability of the processor are the reason they are 
discontinued. The good news is that moving over to the newer generation seems 
to resolve this, but comes with a handful of version 7 quirks.
On 4/11/2023 5:55 PM, Alex Kessler wrote:

Been experiencing this bug for years while running NAT and connection tracking. 
 Rebooting every few months while running v6 latest.  Does v7 have any known 
fixes to resolve these watchdog reboots?





-----------------------------------------------------------------------------------




From: "Colin Stanners" < cstanners at gmail.com >
To: "af" < af at af.afmug.com >
Sent: Monday, December 21, 2020 12:59:09 AM
Subject: Re: [AFMUG] Mikrotik 1072 Frustrations

This last year, I've seen a MikroTik CCR1072 switch from long being rock-solid 
to now having occasional random reboots (from watchdog) or 100% CPU usage, 
which strangles the BGP process. In the latter case, tools->profile would show 
the firewall taking 100% of CPU, even after temporarily disabling all firewall 
filter and NAT rules and connection tracking. Not fun.

MT tech support did not seem super helpful or interested, mostly recommending 
to disable watchdog (unacceptable on a production router) or to upgrade 
firmware (without specifying the suspected cause of the problem or nature of 
the fix).

Tried 1 update, that didn't seem to help, have now tried another...

On Sun, Dec 20, 2020, 11:38 PM Steven Kenney < steve at wavedirect.org > wrote:
MIkrotik has been rock solid for me for years. Until this year and the 1072's. 
Random reboots set off by watchdog timer on all of my 1072's. Some more than 
others. Threads in the forum all discuss the same problem exactly. Its a 
connection tracking issue.. however I need connection tracking on one 
particular router. I've adjusted everything I could. Firmware and board 
firmware all up to date etc. Happens randomly with low levels of traffic, high 
levels of traffic, sometimes a couple times a day, sometimes weeks. No DDOS 
evidence at all from upstream routers. Configs checked and rechecked by third 
party experts. I graph everything about the Mikrotik and there are no clues or 
anything abnormal happening before the crash. Plenty of memory, disk space, CPU 
etc. Replaces all the trannies, power cables and such. Not running BGP only 
OSPF on the one that is giving me the most trouble.

Even have a serial console cable plugged into them to my opengear and set it to 
log pretty much everything to console including the kernel and nothing. A hard 
freeze.

Then there is Mikrotik support... I've never needed their support before until 
now. So I put a ticket in and the shitty attitude I'm getting from them seems 
like they KNOW there is something wrong with the hardware and they are 
intentionally not being helpful. It is pretty clear to see with all the people 
reporting this issue that there IS an issue.

If this is any indication of how things are going to go with Mikrotik on the 
newer hardware going forware I think its time to jump to an enterprise level 
system. Juniper most likely. Shame because they are just about keeping up with 
the demands with their hardware. Getting closer to 100Gbps etc and ROS7 ... but 
at their current pace I think we've outgrew them.

All the threads discussing this issue has been absolutely quiet when it comes 
to Mikrotik jumping in to mention or try to help troubleshoot. I think they 
know they had bad hardware out there and do not want to honor warranties. I've 
heard rumors of bad batches of 1072's.

Anyone else encounter this?

--

Alex
Alex Kessler / TECHNICAL OPERATIONS CENTER
O (Ohio) 740.212.3773 / O (All other markets) 888.966.5690 / 145 Columbus Rd, 
Athens, OH 45701 / point-broadband.com<https://point-broadband.com/>


-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Reply via email to