Hiroshi Nakamura

@bogdanovich Thanks for filing this issue. HTTPClient has retry and it should reconnect I thought. Could you please show a stacktrace for your problem?

Thanks @vincentwoo and @awood, and sorry for the regression. Obviously the document is obviously untrue. Expected behavior for me and OK/NG of 2.7.0.1 implementation. ``` 1. http://foo/bar/baz + path ->...

I fixed the document: https://github.com/nahi/httpclient/commit/f8a4680836dfa8ee23a2382360aee990c9280231 How do you think? EDIT: oops, I introduced unnecessary change in the commit. I reverted and created a new commit here: https://github.com/nahi/httpclient/commit/357df32b87417714b3f1b6219d04e610ca1e9385

According to the error, at that time cookie file contained wrong record(s) in it. It should not happen but I found that saving/loading cookie file is not thread-safe so one...

No idea so far. HTTPClient seems to stall at TCPSocket.new(target) and the underlying CRuby seems to stall at getaddrinfo(3) for getting address information of the specified target, according to the...

@kosh-jelly Thanks. I think the root cause is many CLOSE_WAIT sockets. httpclient should try to reuse connection and close properly but it seems such recycle is not working for you....

@kosh-jelly I think I've understood the situation. Your code uses single HTTPClient instance so connections should be properly recycled, but since you need to access various different domains for fetching...

@kosh-jelly The workaround should work when you apply either of it. I'm pretty sure about the reason of your trouble so you can only try "reset". Hmm, you also need...

Hmm. Actually reset_all is not working for you... CLOSE_WAIT means client haven't call TCPSocket#close but reset_all blocks until it calls TCPSocket#close for all cached sessions. No idea why reset_all doesn't...

Is the error still happens? If true, can anyone provide IIS + NegotiateAuth for me? I want to fix it...