Re: Are prefix+target starting URLs required?

Hello,

Thanks to everyone for the excellent, clearly explained responses.  I figured there had to be a good reason, but I sometimes just need it spelled out for me to buy it.  None of our resources authenticating on EZproxy currently would be impacted by the issue as described, but for consistency’s sake, I will go back and update the URLs on our site.


Thank you!

David

From: Andrew Anderson <[email protected]>
Sent: Friday, December 13, 2019 1:45 AM
To: Noe, David <[email protected]>
Cc: [email protected]
Subject: Re: [EZPROXY-L] Are prefix+target starting URLs required?

Hi David,

ABC-CLIO is another example of a dashed hostname that you may run into.

When I was updating the ProxyPrefix jQuery plugin to add https hostname support, I had asked OCLC how EZproxy handled hostnames with dashes in them.  Unfortunately, I did not receive a sufficiently satisfactory answer to implement the functionality that would "dash" an EZproxy hostname in the same way that EZproxy does based exclusively on the target URL as input.

EZproxy used (and probably still uses) the following embedded hostname for ABC-CLIO when using HTTPS:

www-abc-clio-com.proxy.example.edu<http://www-abc-clio-com.proxy.example.edu>

The algorithm appears to be a simple search and replace of "." with "-" in the hostname, which gives no indication that there was a dash in the original target URL's hostname if there is a need to massage the embedded hostname later.

In contrast, Muse Proxy, uses double dashes when it translates a dashed website name into the embedded hostname format, e.g.:

...www-abc--clio-com.proxy.example.edu<http://www-abc--clio-com.proxy.example.edu>

which makes it quite clear how to translate to or from the original hostname and back.

As it was explained to me at the time by OCLC support, EZproxy maintains an internal mapping that is not available for external programs to use, the format of the proxy-by-host hostnames are undocumented, the embedded hostname algorithm is considered proprietary, and that embedded hostname format should not be relied upon to be consistent between releases.

Now, I have seen EZproxy use the exact same embedded hostname format for over a decade, and I do not anticipate that it will change anytime soon, but since there was no way to cleanly reverse the "dashing" of the hostname, I had to punt and implement a lookup table that ProxyPrefix users on EZproxy could use to add their own mappings by hand instead.

Caveat emptor.

Regarding the use of proxy prefixes, one major benefit to using prefixed links instead of embedded hostname links is that many library platforms have the concept of a proxy prefix and will permit you to use pristine permalink/citation/856$u URLs and have the platform add the prefix for you.  These systems support defining the target URL separately from the proxy server, which means that if your proxy server hostname changes in the future, you simply change the prefix value and the target URL does not have to be updated en masse.  This decouples the target URL from the access platform, so if you move from self-hosted EZproxy to hosted EZproxy or OpenAthens or Muse Proxy, or even back to WAM, you can change the prefix and the rest of your system URL data should not need to change.

Finally, as to migration strategies, we setup the load balancers in front of our Muse Proxy servers to recognize the EZproxy embedded hostname style links and translate them into prefixed links on the fly as the requests traverse the load balancer.  This has been incredibly successful in helping sites that had a large number of embedded hostname links in their library web pages to transition more smoothly to the new platform, and gave them the ability to perform link maintenance as time permits, rather than a frantic rush to find and fix every link before THE BIG DAY(tm) when the migration occurs.

This may be a tactic that you want to explore for your situation as well, as you can also setup rewriting rules for the WAM style embedded hostnames to catch any links that you have not updated prior to the migration date and minimize the impact of links that were not identified prior to the cutover.

Andrew

On Thu, Dec 12, 2019 at 7:57 PM Laszlo Gercsov <[email protected]<mailto:[email protected]>> wrote:
Hi David.

I think you will have a problem if the destination HTTPS URL has "-" hyphens in it. For example, I believe this will work :-

http://yourproxy.uni.edu/login?url=https://macquarie-primoprod.hosted.exlibrisgroup.com

but I think there will be issues with this :-

https://macquarie-primoprod-hosted-exlibrisgroup-com.yourproxy.uni.edu

In the first case, ezproxy would save the correct destination host name, but in the second case the correct hostname would not already be saved by ezproxy, so it would be unreliable to guess which '"-" hyphens to replace with "." dots.

Best,

Laszlo.


________________________________
From: EZproxy List <[email protected]<mailto:[email protected]>> on behalf of Noe, David <[email protected]<mailto:[email protected]>>
Sent: Friday, 13 December 2019 10:43 AM
To: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>>
Subject: [EZPROXY-L] Are prefix+target starting URLs required?

Hello,

I have been at a library using WAM (iii) for a proxy for some time now. We are moving to EZproxy and are configured to proxy by hostname. I would like to clarify a point on best practices.

In the document, "Starting point URLs and config.txt" (help.oclc.org/Library_Management/EZproxy/EZproxy_configuration/Starting_point_URLs_and_config_txt<https://protect-au.mimecast.com/s/hvZhC81Vq2CQDx91Uo28-5?domain=help.oclc.org>), the method for constructing starting URLs is described. Having run EZproxy at a couple of other sites some years ago, this format was familiar to me.

That said, is there a reason to not write starting point URLs in links as they will ultimately be resolved? e.g. https://www-somedb-com.ezproxy.ocls.info<https://protect-au.mimecast.com/s/6Og6C91W8rCxLP8Qi1pwp9?domain=www-somedb-com.ezproxy.ocls.info> ? Is there a reason that URLs necessarily be constructed with a prefix and target? e.g. https://ezproxy.ocls.info/login?url=http://www.somedb.com/<https://protect-au.mimecast.com/s/nRw7C0YKgRs4qljVuLOMXq?domain=ezproxy.ocls.info> ?

I probably misread a document and thought it suggested that this would work fine, so I constructed the URLs as I would have logically expected them to work. This is also how URLs are constructed in WAM on Sierra (and Millennium). It both made sense that it would work like this since we are configured for proxy by hostname and seemed like an easier transition for staff to adjust to EZproxy. My testing would seem to suggest the answer is this is perfectly fine, but this assumption is not supported by the available documentation.

I asked OCLC support and was told that the URL has to be cached in ezproxy.hst or it will not work, and if we were to have to delete this host file and start over, these URLs will not work; however, every time we have added a new stanza, all URLs have worked upon first and continued use throughout this implementation without having ever used the prefix+target construct. Ostensibly, these hosts should not even be in ezproxy.hst until someone has successfully logged in based on this response, right?

If it's the proper way to go, we can certainly adjust our links to the prefix+target format. It would just be an easier transition organizationally if we actually did not have to.


Thank you,

David

********************************************************************
If you wish to stop receiving messages from EZPROXY-L or otherwise amend your preferences,
you can do so <a href="https://www.oclc.org/forms/internet-subscription.en.html<https://protect-au.mimecast.com/s/8ejrCgZ05JfNOn89cKXgtj?domain=oclc.org>">here</a>.
Or email [email protected]<mailto:[email protected]> including the relevant text below in the body of the email:
• To unsubscribe: "unsubscribe EZPROXY-L"
• To receive EZPROXY-L in digest form: "set EZPROXY-L digest"
• To set your options to no mail: "set EZPROXY-L nomail"
• To receive these messages in the future "set EZPROXY-L mail"
To contact the list owners directly please send your message to [email protected]<mailto:[email protected]>.
If you unsubscribe from EZPROXY-L, you will no longer be able to participate in any of its features, including the public forum.
To unsubscribe from all OCLC marketing email communications
(including all OCLC listervs, OCLC Connect emails, OCLC event notifications, product/service/cooperative updates and newsletters),
please email us at [email protected]<mailto:[email protected]>.

________________________________

To unsubscribe from the EZPROXY-L list, click the following link:
http://listserv.oclclists.org/scripts/wa.exe?SUBED1=EZPROXY-L&A=1


--

Andrew Anderson
President & CEO
[email protected]<mailto:[email protected]>
t: +1 (727) 536-0214 x100
f: +1 (727) 530-3126
w: www.lirn.net<https://www.lirn.net/>
Library & Information Resources Network
25400 US Hwy 19 North, Ste. #220
Clearwater, FL 33763
LinkedIn<http://www.linkedin.com/company/library-and-information-resources-network-inc-> | Facebook<http://www.facebook.com/LIRNnotes> | Twitter<http://www.twitter.com/LIRNNotes> | YouTube<http://www.youtube.com/user/LIRNnotes>

****Florida has a very broad Public Records Law. Virtually all written communications to or from State and Local officials are public records available to the public and media upon request. The Orange County Library System policy does not differentiate between personal and business emails. Emails sent via any Library-owned system will be considered public and will only be withheld from disclosure if deemed confidential pursuant to State Law.****

********************************************************************
If you wish to stop receiving messages from EZPROXY-L or otherwise amend your preferences,
you can do so <a href="https://www.oclc.org/forms/internet-subscription.en.html">here</a>.
Or email [email protected] including the relevant text below in the body of the email:
• To unsubscribe: "unsubscribe EZPROXY-L"
• To receive EZPROXY-L in digest form: "set EZPROXY-L digest"
• To set your options to no mail: "set EZPROXY-L nomail"
• To receive these messages in the future "set EZPROXY-L mail"
To contact the list owners directly please send your message to [email protected].
If you unsubscribe from EZPROXY-L, you will no longer be able to participate in any of its features, including the public forum.
To unsubscribe from all OCLC marketing email communications
(including all OCLC listervs, OCLC Connect emails, OCLC event notifications, product/service/cooperative updates and newsletters),
please email us at [email protected].