Open Vugario opened 3 years ago
I can confirm this. The issue for us appears even before doing the request just with:
$client = new GuzzleHttp\Client();
Guzzle 6.5
NewRelic 9.18.1.303
We are facing this issue in our production environment as well 😓
Guzzle: 7.4.0
NewRelic 9.18.1.303
PHP: 8.0.13
Hello @phpfour and @javierwilson , thanks so much for engaging with this issue. Could you include the PHP version used when encountering this issue?
@ZNeumann I've added the PHP version, thanks!
I am also experiencing the same issue:
Same on:
Same here on 7.4.27
Hello all! We just released a new version of the agent (9.19.0) that includes a fix for curl on PHP 8.0. If you could please update your agent and report back whether or not this issue persists? Thanks again for your engagement!
I installed 9.19.0 on a server that hosts our horizon/redis queue and still got this error.
php: 8.0.8, laravel: 8.70.2
@ZNeumann Thanks for releasing the new agent. However, we are still getting the Could not resolve host: router
error.
PHP: 8.1.3
Laravel: 8.83.2
Guzzle: 7.4.1
Any ideas what is the reason? May be we can change something to get around this bug?
https://github.com/newrelic/newrelic-php-agent/releases/tag/v9.18.1.303 May be this is because of route caching? Can somebody test with it (i have only production with new-relic)?
@vismutx we followed the advice here https://stackoverflow.com/a/69631399/1170404 and disabled Newrelic support for Guzzle: newrelic.guzzle.enabled = false
it worked for us but of course, you miss a lot of useful information that Newrelic can no longer provide.
@javierwilson Thanks, Better to miss Guzzle information, than all. because for now we turned off php client at all :)
Currently running into this issue on production.
php: 8.1.7
Guzzle: 7.4.2
Hello, we are now on version 10.5.0; can this be tested against the latest Php Agent version to validate that this issue still exists?
Description
When using the agent version 9.18.1.303 in combination with guzzlehttp 7.3.0 our requests are sometimes changed to have the host "router". We make a simple API call to any url, and it ends up performing a curl request to
router
which in turn throws an exception:cURL error 6: Could not resolve host: router
We noticed this problem in our production environment where we perform many API calls. Most of them succeeded but after a while, some are failing with this described issue. When this error occurs, some API calls still succeed so the error is only happening for a small subset of requests, making this issue very difficult to find and debug.It took us many days to realize this was the cause, but when reverting NewRelic back to its previous release this issue was resolved.
Steps to Reproduce
php:8.0-fpm-alpine3.13
with the new relic agent version9.18.1.303
and GuzzleHttp 7.3.0router
instead of tohttps://foo.com/api/test
Expected Behavior
The request should be sent to
https://foo.com/api/test
instead of torouter
which fails the requestRelevant Logs / Console output
This is our stacktrace, which is no different for the failed requests as for the successful ones (except for the exception handling at the end of course).
Your Environment
We run on GKE with a Docker file based on
php:8.0-fpm-alpine3.13
. GuzzleHttp version 7.3.0 NewRelic agent 9.18.1.303Additional context