Long CONNECT requests are occupying httpd processes
Issue
- Our httpd forward proxy has processes persist with long CONNECT requests in Read states with FIN_WAIT2 sockets. From the logs, we can see requests persisting over 7200 seconds until failing from timeout with a 500:
[24/May/2022:13:58:21 -0700] 127.0.0.1 - - "CONNECT localhost:443 HTTP/1.1" 500 - "Apache-HttpClient/4.5.6 (Java/1.8.0_312)" 7888702753 - TIME: 7888 PID: 113294 139914472724224 49066
[Tue May 24 13:58:46.321159 2022] [proxy_connect:debug] [pid 113294:tid 139914471671552] mod_proxy_connect.c(195): [client 127.0.0.1:52696] AH01019: connecting localhost:443 to localhost:443
[Tue May 24 16:10:23.323963 2022] [proxy:debug] [pid 113294:tid 139914471671552] proxy_util.c(4042): (110)Connection timed out: [client 127.0.0.1:52696] AH03308: ap_proxy_transfer_between_connections: error on client - ap_get_brigade
Environment
- JBoss Core Services (JBCS) httpd 2.4.37 SP11
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.