CCR2216 ros v7.2.3 BGP . All optical interface are hanging after some times.

RouterOS General
1

Hello all,
we are planning to replace one of our CCR1072 with CCR2216 . the 1072 is configured as a border router and have 8 bgp&ibgp sessions.
the routing table have partial routes only approx 60k , the peak aggregate traffic is 25Gbps with 30% cpu usage .

we configured the CCR2216 with all the configuration , as soon as we shift the traffic to ccr2216 . all sessions get established and traffic run normal with cpu usage around 5%. after few minutes , around 20-30 . all the optical interface of the router are hanged and doesnt tx any packets. traffic stop completely , all bgp sessions are down. no device showing in neighbor. although the ethernet interface is fine , we are able to access CCR with it .
we created support file and send to Mikrotik but no luck as of yet. support told us to try disabling the L3 offload , but interface not exchange any traffic even after disabling the hw offload.

if we reboot the router , its start working normally for few minutes and then again same scenario.( we can reproduce the same condition multiple time).

this time we haven’t rebooted the router and its still in interface hanged state , if any of you want us to try some changes we will do and see.

:also , nothing else is configured there , no firewall , no raw table , only routing filters and bgp sessions with few vlans on downlink.
Screenshot (6).png
Screenshot (7).png

2

After a super-backup install 7.3 “stable” and check if do the same.

Form the image, appear a MTU error.
Post the interface config on both sides.

3

Form the image, appear a MTU error.

How can you tell that ?
Because of different l2mtu between qsfp and ether1 port ?

4

Not, for the RX overflow, but on image you see only one side…

5

This is CCR2216

1 R comment=Downlink qsfp1.1 40g name=qsfp28-1-1 default-name=qsfp28-1-1 mtu=1500 l2mtu=1584 mac-address=DC:2C:6E:E0:7F:35 ori
g-mac-address=DC:2C:6E:E0:7F:35 arp=enabled arp-timeout=auto loop-protect=default loop-protect-status=off loop-protect-send-int
erval=5s loop-protect-disable-time=5m auto-negotiation=no advertise= full-duplex=yes tx-flow-control=off rx-flow-control=off sp
eed=40Gbps bandwidth=unlimited/unlimited switch=switch1 fec-mode=auto

This is CRS-326 Downlink switch

 5 RS name=qsfpplus2-1 default-name=qsfpplus2-1 mtu=1500 l2mtu=1584 mac-address=18:FD:74:02:C4:A6 orig-mac-address=18:FD:74:02:C4:A6 arp=enabled arp-timeout=auto loop-protect=default loop-protect-status=off loop-protect-send-interval=5s loop-protect-disable-time=5m auto-negotiation=yes advertise="" full-duplex=yes tx-flow-control=off rx-flow-control=off speed=40Gbps bandwidth=unlimited/unlimited switch=switch1 fec-mode=auto

This is not a mtu issue ,i guess. also half of the L2 ckt terminated in ccr are from telcos. and are working fine in 1072 without doing any changes.

6

We are experiencing the same, the 2216 was working fine for two weeks then no SFP TX traffic with high RX overflow packets. Running 7.10.1.
The only interface that does work is ether1, which is not connected to the switch chip.