UrlConnection timeout required for FeedEntryMessageSource

[pinkeshsagar-harptec]

Expected Behavior

FeedEntryMessageSource should timeout the connection when server is unresponsive for desired amount of time

Current Behavior

FeedEntryMessageSource holds the connection infinitely which is blocking other threads.

Here is stackoverflow url: https://stackoverflow.com/questions/69822045/threads-are-being-blocked-while-receiving-data-from-rss-feed-adapter?noredirect=1#comment123697506_69822045

[artembilan]

It looks like the best way is to expose an UrlConnection as a FeedEntryMessageSource configuration variant.

See the linked SO thread for a workaround.

Will fix this soon.

[garyrussell]

Actually, I don't think we can do that - the UrlConnection would be opened too soon.

Maybe a Supplier<UrlConnection> ?

[artembilan]

Why is that, Gary?
Connection is really done via connect().
Even URL.openConnection() really does not open it, but just create a new instance:

public URLConnection openConnection() throws java.io.IOException {
       return handler.openConnection(this);
}
...
sun.net.www.protocol.http.Handler
...
protected java.net.URLConnection openConnection(URL u, Proxy p)
        throws IOException {
        return new HttpURLConnection(u, p, this);
}

Essentially the same is done in the:

   public XmlReader(final URL url, final Map<String, String> requestHeaders) throws IOException {
        this(url.openConnection(), requestHeaders);
    }

[garyrussell]

Right; but the reader is created at runtime in getFeed() whenever the fetch list is empty - I don't think we can reuse the same connection each time.

[garyrussell]

That said, given that the UrlResource allows customization (as you described on Stack Overflow), do we really need this change at all?

[artembilan]

I don't think we can reuse the same connection each time.

I see your point.

So, yeah... Turning this issue to docs with respective fix in the feed.adoc about possible way to customize.

[pinkeshsagar-harptec]

So does this mean it will be a sequential execution for multiple threads accessing different urls?

[artembilan]

I’m not sure why make such an assumption, but it probably deserves its own SO thread with much more info