From Bugzilla Helper: User-Agent: Mozilla/4.75 [ja] (X11; U; Linux 2.2.17 i686) BuildID: 20000108 Bug 35636 was closed. But I feel this is a still problem. The original author seemed to be concerned about the number of CLOSE_WAIT sockets. My concern is that CLOSE_WAIT sockets never disappear under some conditions. The followings might be the condition. 1. Use HTTP/1.1 2. No cache for the site. 3. Many(some?) images on the site. Reproducible: Always Steps to Reproduce: 1.Clear Mozilla's cache 2.Open http://www.netscape.com/ 3.See sockets' status (e.g. netstat -tcnp on GNU/Linux). Actual Results: While you keep opening the window, the CLOSE_WAIT sockets never disappear. Expected Results: I think Mozilla should close the socket, when the server closes the other side of it. From the server side's view, FIN_WAIT2 sockets remain. They will disappear by the timeout. I think the server keeps useless resources during waiting for the timeout.
Confirmed Platform: PC OS: Windows 98 Mozilla Build: 2001011904 This seems rather silly to let sockets hang around like that and could be used as DoS. Marking NEW.
this is really a regression of bug 35636, and as in that bug, the problem is that while we do cache open socket connections, this problem indicates a leak in the caching mechanism. Socket transport bloat is bug 62388. Marking dupe. *** This bug has been marked as a duplicate of 62388 ***