I am trying to run project on 64-bit Linux host with 32-bit libraries installed as per this thread
https://www.cpdn.org/forum_thread.php?id=8008#54197

but no tasks are being downloaded, despite server status showing over 2000 of them ready to send.
Any particular reason why client is not downloading?

I was able to download one task on 32-bit host installed in VM but running time is rather appalling.

I am trying to run project on 64-bit Linux host with 32-bit libraries installed as per this thread
https://www.cpdn.org/forum_thread.php?id=8008#54197

but no tasks are being downloaded, despite server status showing over 2000 of them ready to send.
Any particular reason why client is not downloading?

I was able to download one task on 32-bit host installed in VM but running time is rather appalling.

With regards to running time, yes these are long tasks

I don't know why you aren't getting any on the 64 bit host though. If you try and download from it before the one hour time out after the last request by hitting the update button, the one hour just starts again. As your computers are hidden, I can't seen any clues there.

What are the messages saying (in the Event Log) ?

What are the messages saying (in the Event Log) ?

Just standard message like this

Tue 29 Oct 2019 20:21:37 GMT | climateprediction.net | Scheduler request completed: got 0 new tasks

However not long after I created this topic I got one task and 2 days later another one. I suspect that server is sometimes picky when it comes to sending new work.[/url]

I’m having the same problem, I have a Ryzen 5 2600 set up (host 1489273) that’s processing quite happily but I set up a second system (host 1495523) that I set up at the end of November with identical software and nothing I’ve tried will persuade it to download any cpdn work units.

Any suggestions?

I’m having the same problem, I have a Ryzen 5 2600 set up (host 1489273) that’s processing quite happily but I set up a second system (host 1495523) that I set up at the end of November with identical software and nothing I’ve tried will persuade it to download any cpdn work units.

Any suggestions?

I had an anomaly recently. Without going into the details, try rebooting and then manually forcing an update. It might work.

EDIT: And I see that you are on BOINC 7.9.3. Version 7.16.3 is better (though not perfect).
You can get it by adding this repository: https://launchpad.net/~costamagnagianfranco/+archive/ubuntu/boinc

I’m having the same problem, I have a Ryzen 5 2600 set up (host 1489273) that’s processing quite happily but I set up a second system (host 1495523) that I set up at the end of November with identical software and nothing I’ve tried will persuade it to download any cpdn work units.

Any suggestions?

I would try again to make sure all the 32bit libraries are installed. For your version of Ubuntu...

sudo apt-get install lib32ncurses5 lib32z1 lib32stdc++-6-dev

If that doesn't do it, post back here with the boinc manager message log messages from when it sends out a request for work and any messages that are a response to that request.

Strange, I had expected one or two messages with regards to emails going out to serial crashers who don't have the 32bit libs. Maybe all my fellow Mods are getting PMs?

Strange, I had expected one or two messages with regards to emails going out to serial crashers who don't have the 32bit libs. Maybe all my fellow Mods are getting PMs?

I've gotten a couple from people who followed the instructions, PM'd me, and I notified Andy who re-enabled their hosts. It might have been they PM'd me since my name's on the posts with the proper instructions for popular distributions.

I’ve checked that the libs specified are loaded which they are but I’m having difficulty adding the repository to update versions, I’ll try again after tea.

Ok, repository added, 7.16.3 installed, system rebooted and a manual update request performed resulting in one work unit downloaded.

Than you for all the assistance - I’ll update the main system in a couple of days when this has bedded in.

It would also be worth a look at the special threads in Number crunching for the HADAM4 models, especially the N216 model.
They require LOTS of memory.

They can also take 20days + to run!

Yes, I now restrict the N216 WUs to one at a time on my main machine and that has 16gb.

I’ll have to look at the second machine, it has 8gb loaded but my account here is only reporting 6gb, could be a problem there.

Yes, I now restrict the N216 WUs to one at a time on my main machine and that has 16gb.

That is plenty of memory for the N216 (they take less than 1.5 GB each).
The real limitation for running multiple work units is the cache on the CPU.
But your Ryzen 2600 is very good at that; I run four at a time on mine without a significant slowdown.

The real question is the OpenIFS, when they come along.

The real question is the OpenIFS, when they come along.

The most demanding of these took a bit over 5GB/core peak in testing. On a machine with 8GB RAM running two at once was OK as they rarely if ever peaked at the same time. Running three or four there was a significant hit because of swapping to disk though total throughput still increased. My ageing desktop with only 4GB of RAM was recognised as not having enough by BOINC and wouldn't download them.

My ageing desktop with only 4GB of RAM was recognised as not having enough by BOINC and wouldn't download them.

I was wondering how they handled insufficient memory. With a little luck, they won't cause problems, though I will be running them only on machines with 32 GB.
I am ready when they are, I think.

Which, apparently, won't be well into next year. :(
But you know what "they're" like with new work. :)

Yes, I now restrict the N216 WUs to one at a time on my main machine and that has 16gb.

That is plenty of memory for the N216 (they take less than 1.5 GB each).
The real limitation for running multiple work units is the cache on the CPU.
But your Ryzen 2600 is very good at that; I run four at a time on mine without a significant slowdown.

The real question is the OpenIFS, when they come along.

Alongside the cpdn WUs I’m running Rosetta and WCG WUs on all cores, again I restrict MIP to one thread at a time.

When OpenIFS comes along I think I’ll need to double the memory to cope.

When OpenIFS comes along I think I’ll need to double the memory to cope.

Probably worth waiting till the next batch or two gets through testing at least whenever that is. I think the later ones peaked about 3.6GB/task but I didn't get any of them so can't be sure.