I found this question while Googling for how to diagnose Amazon Elastic Load Balancers (ELBs) and I want to answer it for anyone else like me who has had this trouble without much guidance.

ELB Properties

ELBs have some interesting properties. For instance:

• ELBs are made up of 1 or more nodes
• These nodes are published as A records for the ELB name
• These nodes can fail, or be shut down, and connections will not be closed gracefully
• It often requires a good relationship with Amazon support ($$$) to get someone to dig into ELB problems

NOTE: Another interesting property but slightly less pertinent is that ELBs were not designed to handle sudden spikes of traffic. They typically require 15 minutes of heavy traffic before they will scale up or they can be pre-warmed on request via a support ticket

Troubleshooting ELBs (manually)

Update: AWS has since migrated all ELBs to use Route 53 for DNS. In addition, all ELBs now have a all.$elb_name record that will return the full list of nodes for the ELB. For example, if your ELB name is elb-123456789.us-east-1.elb.amazonaws.com, then you would get the full list of nodes by doing something like dig all.elb-123456789.us-east-1.elb.amazonaws.com. For IPv6 nodes, all.ipv6.$elb_name also works. In addition, Route 53 is able to return up to 4KB of data still using UDP, so using the +tcp flag may not be necessary.

Knowing this, you can do a little bit of troubleshooting on your own. First, resolve the ELB name to a list of nodes (as A records):

$ dig @ns-942.amazon.com +tcp elb-123456789.us-east-1.elb.amazonaws.com ANY

The tcp flag is suggested as your ELB could have too many records to fit inside of a single UDP packet. I'm also told, but haven't personally confirmed, that Amazon will only display up to 6 nodes unless you perform an ANY query. Running this command will give you output that looks something like this (trimmed for brevity):

;; ANSWER SECTION:
elb-123456789.us-east-1.elb.amazonaws.com. 60 IN SOA ns-942.amazon.com. root.amazon.com. 1376719867 3600 900 7776000 60
elb-123456789.us-east-1.elb.amazonaws.com. 600 IN NS ns-942.amazon.com.
elb-123456789.us-east-1.elb.amazonaws.com. 60 IN A 54.243.63.96
elb-123456789.us-east-1.elb.amazonaws.com. 60 IN A 23.21.73.53

Now, for each of the A records use e.g. curl to test a connection to the ELB. Of course, you also want to isolate your test to just the ELB without connecting to your backends. One final property and little known fact about ELBs:

• The maximum size of the request method (verb) that can be sent through an ELB is 127 characters. Any larger and the ELB will reply with an HTTP 405 - Method not allowed.

This means that we can take advantage of this behavior to test only that the ELB is responding:

$ curl -X $(python -c 'print "A" * 128') -i http://ip.of.individual.node
HTTP/1.1 405 METHOD_NOT_ALLOWED
Content-Length: 0
Connection: Close

If you see HTTP/1.1 405 METHOD_NOT_ALLOWED then the ELB is responding successfully. You might also want to adjust curl's timeouts to values that are acceptable to you.

Troubleshooting ELBs using elbping

Of course, doing this can get pretty tedious so I've built a tool to automate this called elbping. It's available as a ruby gem, so if you have rubygems then you can install it by simply doing:

$ gem install elbping

Now you can run:

$ elbping -c 4 http://elb-123456789.us-east-1.elb.amazonaws.com
Response from 54.243.63.96: code=405 time=210 ms
Response from 23.21.73.53: code=405 time=189 ms
Response from 54.243.63.96: code=405 time=191 ms
Response from 23.21.73.53: code=405 time=188 ms
Response from 54.243.63.96: code=405 time=190 ms
Response from 23.21.73.53: code=405 time=192 ms
Response from 54.243.63.96: code=405 time=187 ms
Response from 23.21.73.53: code=405 time=189 ms
--- 54.243.63.96 statistics ---
4 requests, 4 responses, 0% loss
min/avg/max = 187/163/210 ms
--- 23.21.73.53 statistics ---
4 requests, 4 responses, 0% loss
min/avg/max = 188/189/192 ms
--- total statistics ---
8 requests, 8 responses, 0% loss
min/avg/max = 188/189/192 ms

Remember, if you see code=405 then that means that the ELB is responding.

Next Steps

Whichever method you choose, you will at least know if your ELB's nodes are responding or not. Armed with this knowledge, you can either turn your focus to troubleshooting other parts of your stack or be able to make a pretty reasonable case to AWS that something is wrong.

Hope this helps!

Network load balancing is the distribution of traffic based on network variables, such as IP address and destination ports. It is layer 4 (TCP) and below and is not designed to take into consideration anything at the application layer such as content type, cookie data, custom headers, user location, or the application behavior. It is context-less, caring only about the network-layer information contained within the packets it is directing this way and that.

Application load balancing is the distribution of requests based on multiple variables, from the network layer to the application layer. It is context-aware and can direct requests based on any single variable as easily as it can a combination of variables. Applications are load balanced based on their peculiar behavior and not solely on server (operating system or virtualization layer) information.

The difference between the two is important because network load balancing cannot assure availability of the application. This is because it bases its decisions solely on network and TCP-layer variables and has no awareness of the application at all. Generally a network load balancer will determine “availability” based on the ability of a server to respond to ICMP ping, or to correctly complete the three-way TCP handshake. An application load balancer goes much deeper, and is capable of determining availability based on not only a successful HTTP GET of a particular page but also the verification that the content is as was expected based on the input parameters.

This is also important to note when considering the deployment of multiple applications on the same host sharing IP addresses (virtual hosts in old skool speak). A network load balancer will not differentiate between Application A and Application B when checking availability (indeed it cannot unless ports are different) but an application load balancer will differentiate between the two applications by examining the application layer data available to it. This difference means that a network load balancer may end up sending requests to an application that has crashed or is offline, but an application load balancer will never make that same mistake.

Reference:

Network Load Balancing versus Application Load Balancing