Skip to main content

IPTables Magic

Blog Post after a long long time. Will be trying to write most of the crazy stuffs done in the past 1.5 years at the sad server

This post is going to cover a bunch of hacks done with iptables to improve / make the most out of linux systems' network performance

IPTables Tee
We are building a packet analysis team which does deep inspection of packets and determine anomalies in the system and determine the slowest performing component in the pipeline. Now sending the raw packets to centralized packet analysis system without affecting the performance of the production system is one of our requirements. We decided to use the iptables tee feature which takes a copy of the packet matching the rule and pass it on to the requested gateway in the same network by just changing the mac address on the cloned packet. The original packet follows the normal process

So lets create a similar setup, my laptop is going to forward a copy of http traffic to raspberry pi in the same network. Make sure ip_forward is turned off at the gateway box(box to which copy of packets are forwarded), here raspberry pi. With ip_forward on, we can cause multiple copies of  packets to be routed in network, tcp will kick in congestion control mechanisms seeing retransmissions. Packets are to be blackholed at the raspberry pi, it can do processing there but should not inject it to the network back conflicting with the original packet. IPtable rules at my box for the forwarding
sudo iptables -t mangle -A INPUT -p tcp --sport 80 -j TEE --gateway;sudo iptables -t mangle -A OUTPUT -p tcp --dport 80 -j TEE --gateway
Here is the IP of raspberry pi. At Prerouting chain we route the incoming packet, http response and at the output chain we  route the outgoing packet, http request. When I ran tcpdump in the raspberry pi while opening in my laptop this is the flow I got

Raspberry pi became the sniffer, it sees all http traffic flowing to/from my box

IPTABLES as LoadBalancer

This is completely inspired from Kubernetes infrastructure. Lets run two webservers in raspberry pi at port 81 and 82. Lets add iptable rules to roundrobin traffic between these two webservers in my laptop. VIP will be my laptop's ip( and port 80, Backends will be pi's ip( and port 81/82

iptables -t nat -A PREROUTING -p tcp  --dport 80 -m state --state NEW -m statistic --mode nth --every 2 --packet 0 -j DNAT --to-destination
iptables -t nat -A PREROUTING -p tcp  --dport 80 -m state --state NEW  -j DNAT --to-destination
iptables -t nat -A POSTROUTING -k MASQUERADE

First two rules is to DNAT the packets to the  backend. The caveat is first rule matches only every 2nd packet, those packets wont follow prerouting chain then. Second rule will match all packets which don't match 1st rule
3rd rule is to SNAT so that actual client ip is replaced by laptop's ip so that backend don't respond to client directly


Popular posts from this blog

How we have systematically improved the roads our packets travel to help data imports and exports flourish

This blog post is an account of how we have toiled over the years to improve the throughput of our interDC tunnels. I joined this company around 2012. We were scaling aggressively then. We quickly expanded to 4 DCs with a mixture of AWS and colocation. Our primary DC is connected to all these new DCs via IPSEC tunnels established from SRX. The SRX model we had, had an IPSEC throughput of 350Mbps. Around December 2015 we saturated the SRX. Buying SRX was an option on the table. Buying one with 2Gbps throughput would have cut the story short. The tech team didn't see it happening. I don't have an answer to the question, "Is it worth spending time in solving a problem if a solution is already available out of box?" This project helped us in improving our critical thinking and in experiencing the theoretical network fundamentals on live traffic, but also caused us quite a bit of fatigue due to management overhead. Cutting short the philosophy, lets jump to the story.

Lessons from Memory

Started debugging an issue where Linux started calling OOM reaper despite tons of memory is used as Linux cached pages. My assumption was if there is a memory pressure, cache should shrink and leave way for the application to use. This is the documented and expected behavior. OOM reaper is called when few number of times page allocation has failed consequently. If for example mysql wants to grow its buffer and it asks for a page allocation and if the page allocation fails repeatedly, kernel invokes oom reaper. OOM reaper won't move out pages, it sleeps for some time and sees if kswapd or a program has freed up caches/application pages. If not it will start doing the dirty job of killing applications and freeing up memory. In our mysql setup, mysql is the application using most of the Used Memory, so no other application can free up memory for mysql to use. Cached pages are stored as 2 lists in Linux kernel viz active and inactive. More details here

The server, me and the conversation

We were moving a project from AWS to our co-located DC. We have setup KVMs scheduled by Cloudstack for each of the component in the architecture. The KVMs used local storage. The VMs are provisioned with more than required resources because we have the opinion that in our DC scaling during peak load and then downscaling doesn't offer much benefits financially as we are anyways paying for the hardware in advance and its also powered on. Its going to be idle if not used. Now we found something interesting our latency in co-located DC was 2 times more than in AWS. The time for first byte at our load balancer in aws was 60ms average and at our DC was 112ms. We started our debugging mission, Mission Conquer-AWS. All the servers are newer Dell hardwares. So the initially intuition was virtualisation is causing the issue. Conversation with the Hypervisor We started with CPU optimisation, we started using the host-passthrough mode of CPU in libvirt so VMs dont see QEMU emulated CPUs,