I have a fairly complicated router config running on an 1100AHx4 on version 6.42.9.

When I remove all the devices and just do laptop to laptop, port 4 (bridge) to port 11 (seperate bridge), via a srcnat, and some queue trees, and stream 290mbit of traffic via iperf from one side to another, it works fine. Until I do an snmpwalk of 1.3.6.1.4.1.14988.1.1.3.10.0 (temperature). This causes about 50ms of packet loss (5%), very reliably.

Somewhat less reliably, but still noticable (about 1 in 3 times), when I did an "export compact" I had loss.

As I stripped away parts of the configuration it didn't move much, until I removed the queue tree. These weren't queue drops, and it was still sporadically dropping after removing the queue drops.

Removing an unconnected OSPF instance seemed to drop the outage down to <1%, but it still isn't as clean as I'd expect.

Am I expecting too much from the 1100AHx4?

I'd use a CCR but with a similar config I've seen a lot of packet reorders (upto 700 packets on a 30mbit/3000pps stream). Unfortunately the actual device is in heavy use until Wednesday, to the point that I can't really switch back to the CCR.

Are you watching CPU load during this? When doing just the data, what is CPU load? Does CPU load spike, and hit 100%, when doing the export or SNMP walk?

Maybe do a /export hide-sensitive to give us an idea of 'how' complex the config is, if you have a lot of NAT or firewall rules that must be read through (even if its only matching one) it can significantly increase processing and decrease performance.