[solved] can't ssh AWS EC2 anymore - iptables flushed

Question

Hi,
I killed a AWS's EC2 connexion by flushing iptables on server side, can't connect anymore.
I connect to an Amazon's EC2 instance (with Ubuntu) mainly as a proxy, from a linux system based laptop, to go through an university's firewall.
(For a heads up, I do have some experience with sysadmin on linux, however not that much with network (close to none). I could be qualified as an "aware beginner". As a mean to get some experience with network, a friend of mine, who is most skilled, suggested to use, and helped me to set up an Amazon EC2 instance.
I get a connexion by pulling a ssh tunnel on a terminal:
ssh -i key.pem -D 8080 -p 443 -o 'IdentitiesOnly yes' [email protected]
However for the tricky part,  the university's firewall blocking ssh too.
I could get over it when I eventually setup "stunnel4" to encrypt the whole connection with ssl (found it myself, with some luck in research).
(Installed it when I had the chance to find a proper connexion, which are rare: I found most of hotspot restrictive in that case).
So, it worked well for a while, until I wanted to set up a p2p system (don't laugh guys! It is mostly for study purpose: I needed to down/upload geology maps, which are huge, the firewall always shut our long transfers down, my professors and me).
I tried rtorrent on the server, it worked well.
But because the EC2's 8GB were not big enough, I needed to setup another p2p tool from the laptop (went with Deluge), going through the EC2. For this reason, I opened a range of ports on the Amazon board's security groups (56880-56890, tcp and udp), and added them to iptables, both on ubuntu and laptop, and setup the software accordingly.
# iptables -A INPUT -p tcp --dport 56881:56889 -j ACCEPT
# iptables -A INPUT -p udp --dport 56881:56889 -j ACCEPT
However, because it still did not work, I asked for help to this friend. He told me I set up the software wrong, and suggested me to flush iptables. So, I did it recklessly,  on both laptop and EC2'ubuntu.
# iptables --flush
Of course, doing so killed the connexion to the EC2. There, I could not connect to the server anymore:
ssh: connect to host xx.xxx.xxx.xxx port 443: Connection timed out
Trying to fix it, I found this post, but I faced two difficulties I can not get over with during the process.
1st, on the Amazon EC2 board, after I created a temporary instance, I went to stop the broken instance (i-08091d4b: ok),
detached the broken EBS volume (ok) ,
in order to attach it to the temporary instance (i-64402427: not ok).
I couldn't because the two EBS are supposed to be on the same avaibility zone, which I couldn't setup.
Therefore I couldn't attach the broken EBS on the temp instance.
(I tried then create 2 others temporary instances, in order to check if I could set this avaibility zone).
2nd, I anticipate the fact that the temporary instance doesn't have the "stunnel4" system deployed (software, settings and key), then I won't be able to connect to it anyway.
Is anybody faced similar situation? What can I do to fix the situation?
Last edited by OlaffTheGreat (2014-08-17 23:11:21)

Answer 1

Thanks for your reply.
Unfortunaly, I can't have a physical acces as it is a virtual hard drive somewhere.
As the tuto suggested, I tried also to just reboot the session, hopping the iptables would just reset afterward. It could have looked like what you say about "only partialy resetted the firewall". But it seems really out of order.
Edit: well, it is back online... but I don't know why.
I played randomly with the EBS volumes: detach/reattach them, and when I tried to ssh again to the broken session, I get the connexion...
Last edited by OlaffTheGreat (2014-08-17 23:10:53)