SNS unreachable endpoint
Hi Ben,
Wonder if you've any advice, we moved our sendy installation into a different directory to hide the 'sendy' part, and have thus gone to change AWS SNS endpoints for Bounce and Complaints. The Complaints subscription worked fine but the Bounced one has now been pending all day.
Have you had this issue? Any suggestions or am I at the mercy of Amazon…
Thanks for your help again!
Cheers
Tom
This discussion has been closed.
Comments
Hi Tom,
When you add an SNS subscription, it should immediately be confirmed. Can you double check that your endpoint for bounces is correct? If complaints subscription works, there's no reason why the bounces subscription won't work.
Thanks.
Ben
Yeah, seems weird - i've just been through the logs and found that the amazon checking IP has accessed but is showing an error (on line 51 or bounces.php) which seems to suggest the JSON being wrong?.
I've tried removed and re-adding the topic, then re-subscribing too with no luck.
Planning to send our first proper campaign tomorrow so feeling a bit paranoid...
I'm running the latest version of Sendy 1.1.6.3
Also i've another bug in that our html entities are being stripped if we duplicate the campaign into a new one. Seems to also do it when re-editing. Saving only once leaves them intact though, so for now I can re-paste the HTML in each time I edit.
Sorry this should probably be another thread.…
Hi Tom,
May I know what is your licensed domain so I can do some checks. You're not signed up to the forum with the email used to purchase Sendy.
Thanks.
Ben
Hi Ben,
The domain is http://eml.so, sorry didn't realise i'd used different addresses...
I am using Sendy out of a folder named '-', but that shouldn't break the JSON from Amazon? Have tried a different folder name too but with no luck so reverted back to the dash.
Thanks again
Tom
Hi Tom,
I'm not exactly sure what is wrong though. There's no known issues with bounces and complaints handling. What I noticed is that you're using a modified version of PHP - PHP Version 5.3.3-7+squeeze15. Sendy's tested on regular PHP. So it must be some kind of server environment issue. I have seen one other customer using PHP "squeeze" who faced some other issues as well.
If you'd like to switch servers, eg. Amazon EC2 on a regular LAMP stack (recommended), just PM me.
Thanks.
Ben
Hi Ben,
We had absolutely no problem when we first demo'd Sendy last month on the same server, only difference is we ran 1.1.6.
Squeeze is not a modification, it's standard a Debian release and we develop our own PHP applications on the server and thus can confirm it's totally standard.
Can I downgrade to 1.1.6 again and see if that helps?
Cheers
Tom
Right, this morning I seem to have had more success — i've disabled the firewall on our server and that seems to have allowed the endpoint to register. We've got Amazon's AWS whitelisted in the firewall but it seems that it must use an IP outside the usual range (only sometimes though as the complaints worked yesterday).
Do you know the DNS name of where the requests ping back to? A bit tricky to debug from here.
Sorry to be an ongoing pain!
Cheers
Tom
For anyone else with an outgoing firewall who might have this issue, the DNS name for the endpoint confirmations is 'sns.us-east-1.amazonaws.com' currently.
Cheers
Tom
Hi Tom,
I had no idea you were using a firewall. Good to know that you nailed the issue.
Thanks.
Ben