I did the full update/reboot, and the files uploaded. But this time I lost both CUDA and OpenCL. Faffed around for a while trying to reinstall them - no go. Eventually, simply stopped the client and started it again - GPU was back (I think there's an issue somewhere about needing a delay for the drivers to load, before starting the service). Then had to find a full-feature Manager, because the update had sent me a Debian one with half the menus missing. I think all that's left is to re-enable my backup OpenCL project. And do the other machine...

Uh oh. This does not bode well for when I eventually get round to testing Linux (which I've never used before) to see if it really is 3 times (!) faster at processing than Windows.

Not in general - just the late-lamented special sauce at SETI.

Eh, could someone explain what you're talking about really? My ca-bundle.crt in the Boinc directory (Windows 8.1, BOINC 7.6,.22), is from 2015-04-30, valid from 1998-08-13 and expired 2018-08-14.
It's issued to, and by gte cybertrust global root, and in the certification path tab, it simply say that the cert has either expired, or isn't valid yet.

I have never had any problems at all with any error messages. So, what's this all about really??

Some projects, like Rosetta, started complaining about expired certificates today.

The ca-bundle.crt file is, as the name suggests, a bundle of many different certificates. Microsoft only shows you the details for the one on the top of the pile.

Some projects rely on certificates buried further down in the pile. One of those other certificates expired today, and the projects that rely on it got cut off. We know of LHC, NumberFields, and Rosetta so far - there may be others.

If you are running one of those projects you're in trouble - completed work can't be uploaded, new work can't be issued. But if you're not running one of the affected projects, nothing has changed - their certificates are still in the pile, and still valid.

The second Linux machine is always easier than the first - whichever one you start with. All done, all running. Now for those emails. And is it beer o'clock yet?

The ca-bundle.crt file is, as the name suggests, a bundle of many different certificates. Microsoft only shows you the details for the one on the top of the pile.

Some projects rely on certificates buried further down in the pile. One of those other certificates expired today, and the projects that rely on it got cut off. We know of LHC, NumberFields, and Rosetta so far - there may be others.

If you are running one of those projects you're in trouble - completed work can't be uploaded, new work can't be issued. But if you're not running one of the affected projects, nothing has changed - their certificates are still in the pile, and still valid.

Ah, I see. Now that I look at it in notepad, I can see the other certs. Thanks for explaining Richard. Since I have no problems, I'll leave it as it is.

I've checked my Android devices. Einstein may be one. I have work to report but the servers are down or unreachable.
Rosetta uploads stop with transient errors.
But no trouble with Universe or Asteroids.

Einstein is OK on my Linux and Windows machines.

Edit - my Android tablet @ Einstein has been OK while all this has been going on. Phew.

Now, it's slightly impossible to update the ca-bundle.crt under Android, unless one has rooted their device. So it's very much possible this completely breaks 7.4.53
(And yes, my non connect @Einstein is that other certificate expiration.)

I guess this also means the death of HTC Power to Give (based on BOINC 7.4.39). It was working so very well for me until this point (temperature limits are a bliss)...
Disabling the expired certificates in settings didn't fix the issue for me, editing the ca-bundle.crt also seems really difficult (I need to root my device for that, really?)

I trust Toby most of the time but he might want to add all of our cores to his fleet and rule the world
It was said over there that it was just Win 10 but it is everything (no problem at Einstein)
And over at -dev we have the same thing and can't reload......but funny thing is one of my hosts did get new work when I just left it running to get more work for a while
This is always something that happens on a saturday night usually just VB and this time Sixtrack too.

OK back to watching The Invaders on the 75in 4K UHD

No problems on my two Linux machines on CPDN which uses https. No reports of this problem on the CPDN boards from users though uploading zips to Oz is proving a problem for Windows users at the moment but that is a server in Tasmania that needs size 10 therapy.

No problems on my two Linux machines on CPDN which uses https. No reports of this problem on the CPDN boards from users though uploading zips to Oz is proving a problem for Windows users at the moment but that is a server in Tasmania that needs size 10 therapy.

I've just asked CPDN users to use extra logging detail on that Oz server to decode the "transient HTTP error" - we need to be certain exactly what's going on. The new report yesterday (that you asked for a log for) is within the time bracket for the certificate expiry.

I guess this also means the death of HTC Power to Give (based on BOINC 7.4.39). It was working so very well for me until this point (temperature limits are a bliss)...
Disabling the expired certificates in settings didn't fix the issue for me, editing the ca-bundle.crt also seems really difficult (I need to root my device for that, really?)

The BOINC programs and data directories are under the root data directory, at ./data/data/edu.berkeley.boinc and cannot be looked at or edited without having rooted the device. Total Commander will give an error even when you try to look at ./data, let alone anything deeper.