aws: restore IMDS timeouts after upstream_set in EC2 provider

flb_output_upstream_set() overwrites IMDS client timeouts with output plugin values, causing
credential refresh failures. Restore the correct IMDS timeouts after the upstream is configured.

This patch fixes the timeout issue when output plugins have custom network settings.

Signed-off-by: Eduardo Silva <[email protected]>

Author: edsiper

src/aws/flb_aws_credentials_ec2.c

@@ -49,7 +49,7 @@ struct flb_aws_provider_ec2 {
 
     /* upstream connection to IMDS */
     struct flb_aws_client *client;
-    
+
     /* IMDS interface */
     struct flb_aws_imds *imds_interface;
 };
@@ -171,13 +171,24 @@ void async_fn_ec2(struct flb_aws_provider *provider) {
 
 void upstream_set_fn_ec2(struct flb_aws_provider *provider,
                          struct flb_output_instance *ins) {
+
     struct flb_aws_provider_ec2 *implementation = provider->implementation;
 
     flb_debug("[aws_credentials] upstream_set called on the EC2 provider");
+
     /* Make sure TLS is set to false before setting upstream, then reset it */
     ins->use_tls = FLB_FALSE;
     flb_output_upstream_set(implementation->client->upstream, ins);
     ins->use_tls = FLB_TRUE;
+
+    /*
+     * flb_output_upstream_set() overwrites u->base.net with the instance
+     * network configuration, since EC2 IMDS upstreams require a finite timeout, we need
+     * to restore the values configured during provider creation.
+     */
+    implementation->client->upstream->base.net.connect_timeout = FLB_AWS_IMDS_TIMEOUT;
+    implementation->client->upstream->base.net.io_timeout = FLB_AWS_IMDS_TIMEOUT;
+    implementation->client->upstream->base.net.keepalive = FLB_FALSE;
 }
 
 void destroy_fn_ec2(struct flb_aws_provider *provider) {