Amazon EC2 instances do not appear in Mathworks Admin Center
Show older comments
I am trying to understand how Matlab parallel computing works with multiple "instances" (separate nodes) in the Amazon EC2 environment. So far, I have figured out almost all of the network, permissions, and security issues with Amazon EC2 in order to get Matlab to talk to individual nodes. I can Add a single EC2 node in the Admin Center "hosts" panel. I can see that "reachable" is green and MDCE Service Status is green.
However, I am running into an odd problem now. I started a second EC2 instance as another node, using the same exact AMI setup as the first node. If you are unfamiliar with Amazon EC2, the AMI is essentially a disc image of the boot drive. It contains the operating system and anything that you installed into the AMI. In my case, I installed Matlab into the AMI. So any instance I boot with the same AMI will have Matlab installed.
Now, when I attempt to add the new node to the hosts panel in Admin Center, I receive no errors. But, the new node does not show up in the hosts panel! I run a "Test Connectivity", and everything looks reasonable (there's a few hostname resolution infos, but nothing major). Next, just to see if I can reach the second node on it's own, I start a "New Session" in Admin Center to flush out all the information of the old hosts. I add the second node, and this time it shows up in the Admin Center hosts panel! But when I run the "Test Connectivity", I now get failures in the hostname consistency.
I've attached an image below that shows what is going on. And then below that, I've pasted the logs from the first node Test Connectivity and from the second node Test Connectivity.
Screen shots of the Admin Center in action:
Log output from -138 EC2 instance.
ID # Start Time Stop Time Executing Node Subject Node Test Category Test Description Result Message
1 Dec 3, 2012 3:18:56 AM Dec 3, 2012 3:18:56 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com CLIENT_HOSTNAMES ResolveHostname Test SUCCESS
2 Dec 3, 2012 3:18:56 AM Dec 3, 2012 3:18:56 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST GetClientInfo Test SUCCESS
3 Dec 3, 2012 3:18:56 AM Dec 3, 2012 3:18:56 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST OpenServerSocket Test (on port 27371+) SUCCESS
4 Dec 3, 2012 3:18:57 AM Dec 3, 2012 3:18:57 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST ResolveHostname Test SUCCESS
5 Dec 3, 2012 3:18:57 AM Dec 3, 2012 3:18:57 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com CLIENT_HOSTNAMES ResolveIPToHostname Test INFO Hostname (ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com) and canonical hostname (XXX.XXX.XXX.177) do not match.
6 Dec 3, 2012 3:18:57 AM Dec 3, 2012 3:18:57 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MDCE_CONNECT LockDown Test SUCCESS
7 Dec 3, 2012 3:18:57 AM Dec 3, 2012 3:18:57 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST ResolveIPToHostname Test INFO Hostname (ec2amaz-XXXXXX1) and canonical hostname (XXX.XXX.XXX.177) do not match.
8 Dec 3, 2012 3:18:57 AM Dec 3, 2012 3:18:57 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MDCE_HOSTNAME GetServerInfo Test SUCCESS
9 Dec 3, 2012 3:18:57 AM Dec 3, 2012 3:18:57 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT ResolveHostname Test SUCCESS
10 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:58 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MDCE_HOSTNAME ResolveHostname Test SUCCESS
11 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:58 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT ResolveIPToHostname Test INFO Hostname (ec2amaz-XXXXXX1) and canonical hostname (XXX.XXX.XXX.177) do not match.
12 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:58 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MPI_HOSTNAME ResolveHostname Test SUCCESS
13 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:58 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MDCE_HOSTNAME ResolveIPToHostname Test INFO Hostname (ec2amaz-XXXXXX1) and canonical hostname (XXX.XXX.XXX.177) do not match.
14 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:59 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT PingServerSocketHost Test SUCCESS
15 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:59 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT ConnectToServerSocket Test (remote port 27371) SUCCESS
16 Dec 3, 2012 3:18:58 AM Dec 3, 2012 3:18:59 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MPI_HOSTNAME ResolveIPToHostname Test INFO Hostname (ec2amaz-XXXXXX1) and canonical hostname (XXX.XXX.XXX.177) do not match.
17 Dec 3, 2012 3:18:59 AM Dec 3, 2012 3:18:59 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com PORTS_AVAILABLE CheckServices Test (on port 27350) SUCCESS
18 Dec 3, 2012 3:19:00 AM Dec 3, 2012 3:19:00 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com PORTS_AVAILABLE OpenServerSocket Test (on port 27355+) SUCCESS
19 Dec 3, 2012 3:19:00 AM Dec 3, 2012 3:19:01 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST Cleanup Test SUCCESS
20 Dec 3, 2012 3:19:00 AM Dec 3, 2012 3:19:01 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com PORT_CONNECT PingServerSocketHost Test SUCCESS
21 Dec 3, 2012 3:19:00 AM Dec 3, 2012 3:19:01 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com PORT_CONNECT ConnectToServerSocket Test (remote port 27355) SUCCESS
22 Dec 3, 2012 3:19:01 AM Dec 3, 2012 3:19:02 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com PORTS_AVAILABLE Cleanup Test SUCCESS
23 Dec 3, 2012 3:19:02 AM Dec 3, 2012 3:19:02 AM ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com ec2-XXX-XXX-XXX-138.compute-1.amazonaws.com MDCE_CONNECT Cleanup Test SUCCESS
Log output from -81 EC2 instance.
ID # Start Time Stop Time Executing Node Subject Node Test Category Test Description Result Message
1 Dec 3, 2012 3:21:20 AM Dec 3, 2012 3:21:20 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com CLIENT_HOSTNAMES ResolveHostname Test SUCCESS
2 Dec 3, 2012 3:21:20 AM Dec 3, 2012 3:21:20 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST GetClientInfo Test SUCCESS
3 Dec 3, 2012 3:21:20 AM Dec 3, 2012 3:21:20 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST OpenServerSocket Test (on port 27371+) SUCCESS
4 Dec 3, 2012 3:21:21 AM Dec 3, 2012 3:21:21 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST ResolveHostname Test SUCCESS
5 Dec 3, 2012 3:21:21 AM Dec 3, 2012 3:21:21 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com CLIENT_HOSTNAMES ResolveIPToHostname Test INFO Hostname (ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com) and canonical hostname (domu-12-XXX-XXX-XXX-31-eb.compute-1.internal) do not match.
6 Dec 3, 2012 3:21:21 AM Dec 3, 2012 3:21:21 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MDCE_CONNECT LockDown Test SUCCESS
7 Dec 3, 2012 3:21:21 AM Dec 3, 2012 3:21:21 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST ResolveIPToHostname Test INFO Hostname (ec2amaz-XXXXXX1) and canonical hostname (XXX.XXX.XXX.177) do not match.
8 Dec 3, 2012 3:21:21 AM Dec 3, 2012 3:21:21 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MDCE_HOSTNAME GetServerInfo Test SUCCESS
9 Dec 3, 2012 3:21:21 AM Dec 3, 2012 3:21:21 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT ResolveHostname Test SUCCESS
10 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MDCE_HOSTNAME ResolveHostname Test SUCCESS
11 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT ResolveIPToHostname Test ERROR com.mathworks.toolbox.distcomp.admincenter.testing.infra.util.TestFailureException: No address in list [/XXX.XXX.XXX.177, /0:0:0:0:0:0:0:1, /127.0.0.1, /XXXX:XXXX:XXXX:XXXX:XXXX:XXXX:XXXX:a24e%10, /XXXX:XXXX:XXXX:XXXX:XXXX:XXXX:XXXX:5db1%12, /2001:0:XXXX:XXXX:XXXX:XXXX:XXXX:a24e, /XXXX:XXXX:XXXX:XXXX:XXXX:XXXX:XXXX:2757%11] matched EC2AMAZ-XXXXXX1's address EC2AMAZ-XXXXXX1/10.XXX.XXX.25.
12 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MPI_HOSTNAME ResolveHostname Test SUCCESS
13 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT PingServerSocketHost Test SKIPPED previous test failed: ResolveIPToHostname Test (POOL2CLIENT) (ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com -> client): ERROR
14 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2amaz-XXXXXX1 POOL2CLIENT ConnectToServerSocket Test SKIPPED previous test failed: ResolveIPToHostname Test (POOL2CLIENT) (ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com -> client): ERROR
15 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MDCE_HOSTNAME ResolveIPToHostname Test ERROR com.mathworks.toolbox.distcomp.admincenter.testing.infra.util.TestFailureException: No address in list [/0:0:0:0:0:0:0:1, /127.0.0.1, /fe80:0:0:0:XXXX:XXXX:XXXX:cde6%10, /2001:0:4137:XXXX:XXXX:XXXX:XXXX:cde6, /fe80:0:0:XXXX:XXXX:XXXX:XXXX:3219%12, /fe80:0:0:0:XXXX:XXXX:XXXX:317c%11, /10.XXX.XXX.25] matched EC2AMAZ-XXXXXX1's address EC2AMAZ-XXXXXX1/XXX.XXX.XXX.177.
16 Dec 3, 2012 3:21:22 AM Dec 3, 2012 3:21:22 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MPI_HOSTNAME ResolveIPToHostname Test INFO Hostname (ec2amaz-XXXXXX1) and canonical hostname (10.XXX.XXX.25) do not match.
17 Dec 3, 2012 3:21:23 AM Dec 3, 2012 3:21:23 AM ec2amaz-XXXXXX1 ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com PORTS_AVAILABLE CheckServices Test SKIPPED previous test failed: ResolveIPToHostname Test (MDCE_HOSTNAME) (client -> ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com): ERROR
18 Dec 3, 2012 3:21:23 AM Dec 3, 2012 3:21:23 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com PORTS_AVAILABLE OpenServerSocket Test SKIPPED previous test failed: ResolveIPToHostname Test (MDCE_HOSTNAME) (client -> ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com): ERROR
19 Dec 3, 2012 3:21:23 AM Dec 3, 2012 3:21:23 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com PORT_CONNECT ConnectToServerSocket Test SKIPPED previous test failed: OpenServerSocket Test (PORTS_AVAILABLE) (ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com -> ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com): SKIPPED
20 Dec 3, 2012 3:21:23 AM Dec 3, 2012 3:21:24 AM ec2amaz-XXXXXX1 ec2amaz-XXXXXX1 CLIENT_TEST Cleanup Test SUCCESS
21 Dec 3, 2012 3:21:24 AM Dec 3, 2012 3:21:24 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com PORTS_AVAILABLE Cleanup Test SKIPPED previous test failed: OpenServerSocket Test (on port 27355+) (PORTS_AVAILABLE) (ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com -> ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com): SKIPPED
22 Dec 3, 2012 3:21:23 AM Dec 3, 2012 3:21:24 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com PORT_CONNECT PingServerSocketHost Test SUCCESS
23 Dec 3, 2012 3:21:24 AM Dec 3, 2012 3:21:24 AM ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com ec2-XXX-XXX-XXX-81.compute-1.amazonaws.com MDCE_CONNECT Cleanup Test SUCCESS
I am guessing this is some kind of issue specific to using virtual cloud servers like in Amazon EC2. I had two actual physically separate hosts in my local network to learn / test the Admin Center, and they both showed up separately every time. This issue only started happening when I moved to Amazon EC2 virtual servers. So far, I have yet to ever have 2 separate virtual Amazon EC2 servers show up together in the Admin Center with reachable = green and MDCE Service Status = green. I was able to accomplish this with my local hosts on my local network. The two Amazon EC2 servers are exactly the same Matlab installation, same AMI, same available zone, same virtual machine type, same everything.
It almost feels to me like Admin Center is being confused by the existence of these "virtual" servers. Somewhere in the process of resolving the public IP, local IP, public DNS, or local DNS, Admin Center thinks that these virtual servers are the same physical servers. Therefore, Admin Center does not account for them separate. Instead, updating its LOG files, configurations, or something to essentially say that the same server can be found at 2 separate public IPs or public DNSs.
Has anyone tinkered with Amazon EC2 instances in this manner? Any ideas?
Answers (0)
Categories
Find more on Startup and Shutdown in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!