Given that the error message we noted applied to establishing an SSL / HTTPS connection on port 443, any clues are likely to be found in the Apache logs, rather than BOINC. I don't think any part of BOINC has been reached yet.

Ah, right. Apologies, only half following this thread as away. Doesn't that point to an issue with certificates then?

Am not sure where the httpd server logs are on CPDN, I'll ask Andy.

---

---
CPDN Visiting Scientist

Just to confirm what's supposed to happen:

The BOINC client issues a request to the 'master' url:

https://climateprediction.net/

The intention is to download the entire front page source code, as if for display - but, not being a browser, it can't display anything.

Instead, it searches through the page code, looking for something like

		<!--CPDN BOINC meta tags - MUST be on front page 
			<scheduler>https://www.cpdn.org/cpdnboinc_cgi/cgi</scheduler>
			-->
		       <link rel="boinc_scheduler" href="https://www.cpdn.org/cpdnboinc_cgi/cgi"/>

- that's ours. Other projects use something similar, like

    <!-- <scheduler>https://numberfields.asu.edu/NumberFields_cgi/cgi</scheduler> -->
<link rel="boinc_scheduler" href="https://numberfields.asu.edu/NumberFields_cgi/cgi">

I'm wondering if there's some sort of anti-malware or firewall operating which sometimes says "you want our front page, but you're not a browser - that sounds phishy to me" and blocks it. edit - like an over-zealous 'no robots' tag, or similar.

Just to confirm what's supposed to happen:

Just to add, if BOINC client manages to download the master url, you'll have a file called master_climateprediction.net.xml with this information in most commonly c:\programdata\boinc directory (if windows).
Meaning, if you do not have this file, or the file only include a warning or something, the problem is to download https://climateprediction.net

If on the other hand you do have a "valid" master_climateprediction.net.xml, the problem is to connect to the scheduling server.

Eastern Time Zone, East Coast USA.

I can not get anything from the BOINC server or web site.

If I try to connect I am simply told it failed. No files are downloaded.

Sorry, I can access the web site (this site) just fine on my computer using my browser. I meant via BOINC I can't access the web site/server.

Just a thought, is there a way to download the master_climateprediction.net.xml file and place it in the projects directory and more to the point, would this help?

Edit: or indeed copy it from a machine that has the file.

I wondered about that. My browser (Chrome, for daily use - I haven't checked Firefox) allows me to 'display source', and from there, 'save as'. If Skillz can reach climateprediction.net with a browser, he should be able to save it to disk with the name and extension changes noted. (and if he can't, that would be useful to know, too).

It might not work - I think BOINC has a habit of opening a blank file to hold the data, before requesting the download. That would delete the manually-downloaded file, unless it's protected by being marked as read-only. And I don't know of any way of forcing BOINC to read the master file, except in the context of the attachment process.

I know when I clicked on the file to open it from the projects directory, it wouldn't let me open it in a browser though I could if I copied it from there to my downloads folder where it told me the file was badly formed or some such. Opening in a text editor was no problem.

Depends on the OS. In Windows, the OS will expect a browser file to have .htm or .html as the 'extension' at the end of the name. BOINC saves it with the extension .xml, but that doesn't work either, because it doesn't contain XML formatted text. Hence, only a plain-text editor (step forward Notepad!) will show you what's really going on.

8/18/2024 8:16:44 PM | climateprediction.net | [http] [ID#1] Info:    Trying 129.67.241.50:443...
8/18/2024 8:16:44 PM | climateprediction.net | [http] [ID#1] Info:  Connected to climateprediction.net (129.67.241.50) port 443 (#2006)
8/18/2024 8:16:44 PM | climateprediction.net | [http] [ID#1] Info:  schannel: disabled automatic use of client certificate
8/18/2024 8:16:44 PM | climateprediction.net | [http] [ID#1] Info:  ALPN: offers http/1.1
8/18/2024 8:16:45 PM | climateprediction.net | [http] [ID#1] Info:  schannel: next InitializeSecurityContext failed: Unknown error (0x80092013) - The revocation function was unable to check revocation because the revocation server was offline.
8/18/2024 8:16:45 PM | climateprediction.net | [http] [ID#1] Info:  Closing connection 2006
8/18/2024 8:16:45 PM | climateprediction.net | [http] HTTP error: SSL connect error
8/18/2024 8:16:46 PM |  | Project communication failed: attempting access to reference site
8/18/2024 8:16:46 PM |  | [http] HTTP_OP::init_get(): https://www.google.com/
8/18/2024 8:16:46 PM |  | [http] [ID#0] Info:    Trying 74.125.21.106:443...
8/18/2024 8:16:46 PM |  | [http] [ID#0] Info:  Connected to www.google.com (74.125.21.106) port 443 (#2007)
8/18/2024 8:16:46 PM |  | [http] [ID#0] Info:  schannel: disabled automatic use of client certificate
8/18/2024 8:16:46 PM |  | [http] [ID#0] Info:  ALPN: offers http/1.1
8/18/2024 8:16:46 PM |  | [http] [ID#0] Info:  ALPN: server accepted http/1.1

The Schannel configuraiton: disabled automatic use of client certificate, isn't causing issues when connecting to a reference site (google.com) but seems to when trying to connect to climateprediction.net. I was able to attach an old laptop using instructions from View Account Key in the Account Information, had http_debug on and that schannel message wasn't there. It does seem like some kind of local configuration, done by the user or perhaps some networking filtering or similar programs.

Here's a description of that message: https://brainly.com/question/43026691

And I don't know of any way of forcing BOINC to read the master file, except in the context of the attachment process.

If BOINC fails 10 times in a row to successfully connect to scheduling server then master file is tried re-downloaded.

But did you manage to complete the next two stages after that - fetching the master file, and locating the scheduler?

The BOINC coding change that enabled schannel is a bit more helpful in this case.

https://github.com/BOINC/boinc/pull/4545

Build it with schannel instead that allows curl to use Windows certificate store and doesn't relay on ca-bundle.crt anymore.

You'll see that I was assigned as [sole] tester for that change, but wasn't able to comply until a VS2019 compiled version was available.

Note that I dropped out of the conversation part-way through - the next bit was probably a private California conversation while I was asleep. I came back the next day, and eventually gave in.

Note also that CPDN gets a special mention in the testing notes, because it was struggling with the University infrastructure at the time I was trying to test.

And a final note - there's no mention of this in the client release notes for BOINC. Judging by the dates, it's probably been around since v7.20.

That suggests another possibility for Skillz - try installing an older version of BOINC first (I'd suggest v7.16.11 or earlier - the next Windows build wasn't until v7.16.20, over a year later. You'll need an updated ca-bundle.crt file, but we can help you find that). Then, once attached, upgrade to a newer version.

But did you manage to complete the next two stages after that - fetching the master file, and locating the scheduler?

I think so, couple of the things it says in the Event Log are "Master file download succeeded" and "Scheduler request completed". Also master_climateprediction.net.xml exists as well as sched_request... and sched_reply... xml files with what look like normal entries.

Would it be helpful info for Skillz to open PowerShell and run the following command and post up to first 30 lines?

curl -v https://climateprediction.net

Don't know what changed, but I was able to attach a local host to CPDN and got two tasks. No clue what changed. I just decided to give it another shot today and it worked.

I just decided to give it another shot today and it worked.

Glad it is now working for you even if it does mean we have to wait till the problem occurs for someone else before we can track down the root of the problem.
I am guessing Andy and Glenn will have the location of the relevant logs to hand allowing them to be saved for checking another time.