search

donald-jackson / kannel-pluginbox

Kannel Pluginbox allows users to implement their own business logic for routing, billing, etc in Kannel.
Other
12 stars 9 forks source link

'Couldn't fetch' #6

Closed jompigrande closed 4 years ago

jompigrande commented 7 years ago

Hello Donald, I figure out that im having one random 'Couldn't fetch' error daily on every site, i dont have debug enabled to tell much more about this, but the weird thing it's that apache reports 200 as it was successful called the http processor:

grep '2017-01-12 17:37:54' /var/log/kannel/kannel-pluginbox.log.1 2017-01-12 17:37:54 [32047] [12] ERROR: Couldn't fetch http://127.0.0.1/XXXXXX/http-processor.php 2017-01-12 17:37:54 [32047] [5] WARNING: [HTTP-PLUGIN] Upstream URL (null) rejected msg with status -1 (outside >= 200 < 400), rejecting 2017-01-12 17:37:54 [32047] [5] WARNING: Plugin chain has asked us to reject this message

grep '17:37:54' /var/log/apache2/access.log 127.0.0.1 - - [08/Jan/2017:17:37:54 -0600] "GET /XXXXXXX/http-processor.php HTTP/1.1" 200 245 "-" "-"

What did you think that can be happening, let me know if its better to enable the debug on one of the servers to give more details about this.

Regards, Jose

donald-jackson commented 7 years ago

Hi Jose, how heavy was the load at this time (how many messages/sec were you processing) ?

jompigrande commented 7 years ago

Today i have one peak of 20K messages per minute distributed in 6 servers, and it didnt throw any 'Couldn't fetch', i check graphs from 08/Jan/2017 and doesnt have that peaks at that hour, i think i will need to rise debug level and check what its happening when the error cames.

donald-jackson commented 7 years ago

I would suggest checking in your PHP code/Apache, I am 99.9% sure this is an issue related to your web server servicing the request. Although the timestamp you provided is the same it could be for another transaction as there are multiple in a single second.