aledsage
commented
11 years ago So questions are:
- how to avoid this.
- how to detect such problems and work around it.
Problem is that the thread stack shows it was still executing the apt-get install. We wouldn't know that it had failed - perhaps it was just being very slow, and the ssh connection disconnecting caused it to get into the "dpkg was interrupted" state.
So if we lose connection during such an operation, do we have enough retry logic (we have some at the SshjTool level), should we check for this error state (and try remediating action), and/or should we discard this VM and retry with a new VM?
Short term, we should ensure OpenGamma demo runs on something bigger than t1.micro in AWS.
When trying to run the ActiveMQ entity in aws-ec2 us-east-1, running Ubuntu 13.04 on a t1.micro. (I know, t1.micro is crazy small - but this failed before even getting to ActiveMQ)...
It hung when trying to run
apt-get install -y --allow-unauthenticated openjdk-6-jdk.The problem may be related to: http://askubuntu.com/questions/163200/e-dpkg-was-interrupted-run-sudo-dpkg-configure-a
After Duncan had terminated his ssh session (disconnecting from network), I logged in and looked at the box.
dpkg, my ssh session closed each time I ran it!The brooklyn web-console showed:
duncan@ip-10-202-43-188:~$ sudo dpkg --get-selections | grep jdk
sudo apt-get check
duncan@ip-10-202-43-188:~$ sudo dpkg --configure -a
duncan@ip-10-202-43-188:~$ sudo dpkg --configure -a
duncan@ip-10-202-43-188:~$ java -version
duncan@ip-10-202-43-188:~$ sudo apt-get install -f
duncan@ip-10-202-43-188:~$ sudo dpkg --configure -a