chris

after several hours of poking around and turning things off and on again I found the issue. Proxmox, for some reason I am yet to discover, even though it had 0 VMs running and was effectively a clean install, had 4091 out of DD-WRT's 4096 maximum concurrent connections open, I didnt find out where the heck the connections were to or why the heck it was doing that, but after I rebooted it the router hasnt crashed despite me putting 150mbps through it. Seems to be fine now. Why the heck would proxmox even use that many connections?

pptpd, had no devices connected

Time: 00:19:21 up 19 min, load average: 2.12, 2.16, 1.56 /var/log/messages on DD-WRT shows nothing.

nope, still getting 99-100% sirq at seemingly random times which effectively prevent network access. Also happened when the router was disconnected from the modem, so I'm not sure what's causing it.

ok so I installed DD-WRT, and it still spiked once, but as far as i can tell the issue has gone away, for now at least.

all static, DHCP is handled by the router regardless of which AP it's connected to.

gonna try flashing dd-wrt

dmesg | tail -200:xt_TCPMSS: bad length (1047 bytes)printk: 45952 messages suppressed.xt_TCPMSS: bad length (1047 bytes)printk: 45995 messages suppressed.xt_TCPMSS: bad length (1047 bytes)printk: 45950 messages suppressed.(same thing over and over) it would appear something is causing a shitload of CPU utilisation whenever I get packet loss, and xt_TCPMSS: bad length shows up a LOT in dmesg

from what i can glean from google, sirq is using 83% CPU usage to handle ethernet interrupts, which would explain why it has such high latency over ethernet. Anyone got any ideas about how to fix this?

CPU: 0.0% usr 16.6% sys 0.0% nic 0.0% idle 0.0% io 0.0% irq 83.3% sirq also I recommend you read this guide about load averages: http://www.howtogeek.com/194642/understanding-the-load-average-on-linux-and-other-unix-like-systems/ 1.35, -> system overloaded by 35% average over past 1 minute 1.31, -> system overloaded by 31% average over past 5 minutes 1.18 -> system overloaded by 18% average over paste 15 minutes so the router is overloaded. nevertheless, here's the traceroute you wanted $ tracert google.comTracing route to google.com [62.253.72.148]over a maximum of 30 hops: 1 54 ms 56 ms 57 ms NSACollectionHu [192.168.1.1] 2 * 130 ms 118 ms cpc4-enfi18-2-0-gw.20-2.cable.virginm.net [77.99.180.1] 3 126 ms * * hari-core-2a-ae7-627.network.virginmedia.net [81.96.18.61] 4 133 ms 123 ms 125 ms brnt-bb-1a-ae8-0.network.virginmedia.net [81.96.16.201] 5 126 ms 121 ms 117 ms brnt-icdn-1-ae0-0.network.virginmedia.net [62.254.84.18] 6 * 16 ms 19 ms 148.72-253-62.static.virginmediabusiness.co.uk [62.253.72.148]Trace complete.

nope, using standard firmware (latest version available for this revision) Firmware Version: 1.1.29.162351Firmware Builddate: 2014-07-16 10:59Product.type: productionLinux: Linux version 2.6.22 (root@build-vm) (gcc version 4.2.3) #1 PREEMPT Wed Jul 16 03:00:48 PDT 2014Board: carrera

Also to note: lighttpd is running on port 10000 on this router, which i hadnt picked up before, all it shows is 403: Forbidden though

C:\Users\whyrlpool>tracert 192.168.1.44Tracing route to vpn [192.168.1.44]over a maximum of 30 hops: 1 <1 ms <1 ms <1 ms vpn [192.168.1.44]Trace complete.

So after some poking around, it appears my router is at 100% CPU usage top -bn1Mem: 76880K used, 47476K free, 0K shrd, 9068K buff, 30480K cachedCPU: 0.0% usr 16.6% sys 0.0% nic 0.0% idle 0.0% io 0.0% irq 83.3% sirqLoad average: 1.40 1.33 1.19 2/84 18773 PID PPID USER STAT VSZ %MEM CPU %CPU COMMAND 4 2 root RW< 0 0.0 0 94.1 [events/0]18773 18674 root R 1928 1.5 0 5.8 top -bn1 5544 5540 root S < 8676 6.9 0 0.0 /usr/bin/guardian 5434 1 root S < 8676 6.9 0 0.0 /usr/bin/guardian 5543 5540 root S < 8676 6.9 0 0.0 /usr/bin/guardian 5547 5540 root S < 8676 6.9 0 0.0 /usr/bin/guardian 5545 5540 root S < 8676 6.9 0 0.0 /usr/bin/guardian 5546 5540 root S < 8676 6.9 0 0.0 /usr/bin/guardian 5540 5434 root S < 8676 6.9 0 0.0 /usr/bin/guardian 4831 1 root S 8248 6.6 0 0.0 devidentd -r /tmp/devregex.json -p 4866 1 root S 7260 5.8 0 0.0 lighttpd -f /etc/lighttpd.conf UpTime: 14:09:11 up 1:18, load average: 1.35, 1.31, 1.18 QoS is off, and i'm not sure whats causing this

two web browsers, no p2p, nothing saturating the network. The switch is a 5 port ProSafe.