Background requests going to the correct IP but wrong Host.

It caused too many issues for us with no real benefits, so we no longer include it. There is nothing related/forced in the BOA configuration, so you could try to use it as any other module, if you wish.

Oops, it seems that we now force it as disabled after too serious issues reported, so you would need to modify /var/xdrago/usage.sh script to avoid having it disabled every morning.

I think that we shouldn't force it as disabled for everybody.

Fixed in HEAD and stable: http://drupalcode.org/project/octopus.git/commit/86cbe8b

Just curious, what are some of the reported issues with httprl? I can't fix what I don't know :)

need more info for this bug report

Closing issue as no new info has been given. Re-open if you have more info.

I have requested some feedback/details from our users who reported the "issues" in the past, but I guess it will take some time to collect these reports, sorry for the delay here.

Thanks for the update :)
I'll move this back to postponed.

I would also like to hear what type of problems are in httprl.

It's been almost a month. Any news on what might have been the reported issue?

I would love to submit complete report with steps to reproduce issues, but I guess it is not the best time in the year to expect fast response from people, and so far I have received only reports about redirect loops and sending requests to the wrong vhost in Nginx etc. I still need more details to determine if it was really the module issue, Nginx config fault, user side wrong configuration or something else. I need some steps to reproduce the problem, so I can offer better feedback.

Based off the limited information given I think I found the bug. The report would have to do with a multisite and/or multiple drupal installs behind one IP. When issuing a background request and the httprl_server_addr variable is set to an IP the host is also set to that IP. Thus who ever gets served when no hostname is set would get a lot of requests.

The following patch has been committed. Setting to active as I need to do some followup to this.

That sounds similar to the other issue we have had with AdvAgg in the past: #1142990: Why SERVER_ADDR is used instead of HTTP_HOST to build $url in function advagg_css_js_file_builder() ?

I will give it a try. Thanks!

Codewise I thought there would be more to do in this case. Pretty sure the patch in #14 covers everything. I've revamped the readme though and its usage of httprl_build_url_self().

The following patch has been committed.