Why do my Google App Engine SDK EC2 instances keep dying?
I've setup the Google App Engine SDK EC2 instance several (actually perhaps twenty) times now (both using the tools here and also by directly firing up a blank Fedora 8 instance on EC2 and installing the SDK etc.) and everytime - everytime - no matter how I go about it, the server works fine for a few minutes and then just simply becomes unavailable after 5-10 minutes. It stops responding to anything, ssh, the port 2999 console, the demo app on port 8080, everything - it just pretends to not exist.
This is has happened a few times even before I can upload my own application to test it - so it's not something I've written (and anyway my code is working just fine on my localhost - and this problem has not appeared when using localhost - only when I want to deploy something to the web). It's pretty bizarre, and incredibly frustrating as I've now been stuck on this for 27 hours and counting!
Ideas? Thoughts? Miracle solutions anyone? Anyone else encounter this problem?
Edit: I forgot to say that few weeks ago I ran an App Engine SDK instance on EC2 for a week without issues. Pretty sure I'm not doing anything different now. Using the same machine images etc. That experience just makes it even weirder...
This is has happened a few times even before I can upload my own application to test it - so it's not something I've written (and anyway my code is working just fine on my localhost - and this problem has not appeared when using localhost - only when I want to deploy something to the web). It's pretty bizarre, and incredibly frustrating as I've now been stuck on this for 27 hours and counting!
Ideas? Thoughts? Miracle solutions anyone? Anyone else encounter this problem?
Edit: I forgot to say that few weeks ago I ran an App Engine SDK instance on EC2 for a week without issues. Pretty sure I'm not doing anything different now. Using the same machine images etc. That experience just makes it even weirder...
1
person has this question
I have this question, too!
Tell me when someone answers.
The more people who ask this question, the more it gets noticed.
The more people who ask this question, the more it gets noticed.
Create a customer community for your own organization
Plans starting at $19/month
-
Inappropriate?Strange.
Ok, given that it is happening with the Fedora template at EC2 also, it makes me wonder if it is Elastic Sever related. BUT...let's use Elastic Server as a collaborative platform to figure this out! (at least try)
Ok - here are some hypotheses to test.
a) There is something ambient going on around your network where access gets lost.
b) There is something wrong with Elastic Servers created by you - but it is something to do with your On-Demand account/setup.
c) There is something wrong with elastic servers created with the App Engine SDK
d) There is something wrong with your Amazon EC2 credentials or setup at Amazon.
a - TEST: Please use ESOD to build a new instance of the machine you have been trying with your EC2 credentials and deploy/launch at EC2. Email me at pjkerpanDUDE at cohesiveft.com with the appropriate @ sign and remove the DUDE from my name - forward me the email you receive from Elastic Server On-Demand with the URL and SSH credentials to your machine. Then we let it soak at EC2 and we will periodically check access to it.
b - TEST After testing a, then please make an Elastic Server with just Jboss in it and deploy/launch to EC2. Forward me the email. This will give us a port 2999 (our Elastic Server Manager), port 22 (ssh) and port 80xx (JBOSS) device. Let's see if that runs ok for a while.
c - TEST After testing b, then please make a Server with App Engine in it and deploy/launch to EC2. Forward me the email and we will poke around nside of it. See if we see anything strange.
d - TEST If both you and us are losing contact with the server created in A and B. Then we will jump to this test. You will then let me know your ESOD username (NOT PASSWORD). I will share MY EC2 credentials with you. Then build just a plain server, nothing in it, using my credentials to EC2. Then will plan from there depending on the result.
Or take some of this "offline" by emailing me at the address above in test A.
The company says
this answers the question
-
Thanks for that, I will set up those tests now and let you know. Yep - it's a strange one! -
By the way, I set up (for a separate purpose) a Ubuntu Hardy base EC2 image about 20 hours ago - and no problems with it so far, I think that helps to narrow it down to something with the App Engine ami or the Fedora Core 8 base ami. See how it goes.
Loading Profile...



EMPLOYEE