If I publish a pre-print paper on arXiv, how can I guarantee exclusive rights to the publisher afterwards? Am I unable to publish on non-open access journals after I publish a pre-print on arXiv ?
You are generally allowed to publish even in a non-open access journal even if a pre-print is on the arXiv. Most journal copyright agreements explicitly allow the authors to post the article online. Here's an example of a fairly generous one:
The ASL hereby grants to the Author the non-exclusive right to reproduce the Article, to create derivative works based upon the Article, and to distribute and display the Article and any such derivative work by any means and in any media, provided the provisions of clause (3) below are met. The Author may sub-license any publisher or other third party to exercise those rights.
and a less generous one which still allows the author to post a copy online:
I understand that I retain or am hereby granted (without the need to obtain further permission) rights to use certain versions of the Article for certain scholarly purposes, as described and defined below (“Retained Rights”), and that no rights in patents, trademarks or other intellectual property rights are transferred to the journal.
The Retained Rights include the right to use the Pre-print or Accepted Authors Manuscript for Personal Use, Internal Institutional Use and for Scholarly Posting; and the Published Journal Article for Personal Use and Internal Institutional Use.
I've seen examples where the journal actually did some genuine copyediting beyond what the referee did where the author wasn't allowed to post the version that benefited from the copyediting, but could still post the earlier version.
So, for most journals, the answer is that you're allowed to post the article online because it's specifically allowed by the document they ask you to sign. But it is possible that posting on the arxiv will rule out particular journals that have more restrictive policies.
Not necessarily, but this is to be dealt with on a case-by-case basis. You can have an idea of which policies have been adopted by which publishers/journals by having a look at this webpage.
You should ask senior people in your field, or look at the polies and publishing agreements of journals you care about.
In math, every major journal will accept submissions of papers that have previously been posted to the arXiv. (I'm pretty sure that every journal will, but I suppose I can't rule out some obscure exception. If any journal tried to enforce a policy against submitting papers that were on the arXiv, there would be a big uproar and they would have no choice but to allow it.) I believe the same is true in physics and CS.
But keep in mind that this may vary between fields. For example, my understanding is that the American Chemical Society has particularly draconian copyright and dissemination policies, and that they may object to arXiv posting. They are on the wrong side of history, but it doesn't mean they can't still cause trouble. And then there are journals like Science, with embargo policies and corresponding rules about what constitutes prior distribution.
So the answer is that this is very simple if you're in a field in which the arXiv has become widespread, but much more subtle otherwise.
The Ingelfinger rule is generally disregarded now by scientific publishers. F1000 has done a lot of research into this for their poster repository. Of the few notable publishers who do suggest they do not allow preprints, blood was singled out as apparently they said they wouldn't allow it but appeared to have no way of checking this and so do allow it, albeit by default. The following 2 links from the embargo watch blog describe the story:
A standard workaround is that you post a draft of your paper on the arxiv and assign the copyright in the final version to the journal. You can't update the arxiv with your final version, but the journal publishers will only be able to enforce their copyright on those parts of the final version which are not already present in the arxiv prior art.
Since you will almost inevitably make some revisions as part of the submission process, you don't have to explicitly plan to leave anything out of the arxiv version.
Although most journals allow publication of the pre-print (before peer review version) elsewhere, they only do so after they accepted and published the final version of the paper. Otherwise they might take it as previous publication and reject the manuscript. Russel O'Connor ran into this problem with the ACM when he tried to secure open access and free use by putting the material into public domain by uploading it to the ArXiv before submitting it to the ACM.