Still unable to download PDF with Chrome connector

erikmarsh

I have the identical problem as in this thread: https://forums.zotero.org/discussion/70939/unable-to-download-pdf-with-chrome-connector/p1

I've tested the same URL (this time using Berkeley's proxy):
https://www-nature-com.libproxy.berkeley.edu/articles/s41593-018-0104-6
and
https://www-sciencedirect-com.libproxy.berkeley.edu/science/article/pii/S1040618298000536
The URL dots switch to hyphens but that shouldn't matter (I have this option checked in the proxy settings).

I can't get the connector to download any PDF through the proxy (neither from the article webpage or the PDF's page). I also tried right clicking the extension and using "Save to Zotero", but no luck. I have seen three ways it fails: the download hangs (stays grayed out), I get the red X, or the message "An error occurred while saving this item". But I never get the PDF. The citation does import successfully.

The debug ID is D773877039.

I have only one proxy (no "http" prefix): %h.libproxy.berkeley.edu/%p
I'm on a Mac and there are no proxy protocols active in the system prefs.

Any suggestions are welcome! thanks
Erik

dstillman

Can you provide a Debug ID from Zotero (rather than the Connector) for a save attempt that fails?

erikmarsh

Oops. It's D1395252392. That includes four failed attempts (for the first link above).

adomasven

(3)(+0001046): Getting contents of /Users/[username]/Zotero/storage/tmp-bWhmfX/Fleming et al. - 2018 - Neural mediators of changes of mind about perceptu.pdf

(2)(+0000004): Error:
Unix error 2 during operation open on file /Users/[username]/Zotero/storage/tmp-bWhmfX/Fleming et al. - 2018 - Neural mediators of changes of mind about perceptu.pdf (No such file or directory)

This is either security software preventing Zotero from saving pdf files or Zotero not having sufficient permissions to write in the data directory.

erikmarsh

I don't use security software and Zotero *should* have sufficient permissions. I'm no programmer, but to me it looks like Zotero can't find the file, which makes sense, because it never downloaded.

I have been able to use the Chrome Connector to download PDFs from non-proxied sites, so I'm pretty sure the issue is with the proxy.

adomasven

Actually, there's also a bunch of

[JavaScript Error: "www-nature-com.libproxy.berkeley.edu:443 uses an invalid security certificate.

The certificate is not trusted because the issuer certificate is unknown.
The server might not be sending the appropriate intermediate certificates.
An additional root certificate may need to be imported.

Error code: SEC_ERROR_UNKNOWN_ISSUER
"]

in that log, which would suggest why downloads via your proxy fail. Is the proxy certificate valid when you open it via Chrome or did you have to add an exception for it?

dstillman

The Berkeley proxy server is misconfigured, and has been for years.

@erikmarsh, you should send this documentation page (Option 2) to your IT department and ask them to fix it. As explained there, you can verify that this is broken using a standard industry tool, so this isn't about Zotero specifically (though some programs like Chrome work around it).

erikmarsh

Hi Dan – thanks for the links. This was exactly my issue. I was able to solve the problem with your "spaces.internet2.edu" workaround. I don't get how that works but it does. Thanks – now Zotero is much more awesome.

Berkeley switched to EZproxy last year (http://guides.lib.berkeley.edu/ezproxy), a big improvement on their previous proxy. It now gets a B (instead of an F) on ssllabs.com but still has the same crucial issue: an incomplete certificate chain. I submitted a ticket with them and we'll see what happens. There certainly are plenty of Berkeley folks who use Zotero!

pmricks

I recently had the same issue at my university (Vanderbilt) and contacted IT, sending them the documentation page from the SSL server test. They quickly corrected the problem with the intermediate certificate, and the connector started working correctly.