It is set to 10 seconds. What are you seeing that makes you think that it is an hour?

If it helps any, when the server does not send the "Beta - The Clean Energy Project - Phase 2 is not available for your type of computer" message, then the delay requested is the normal 11 seconds.

Only seem to get the 3600 second delay request upon that server status message.

Hutch

i looked back through the log of the 3 computers i have running and do not see any waits from WCG.

noticing the projects you have selected in your post hutch...
No work is available for Help Fight Childhood Cancer
No work is available for Discovering Dengue Drugs - Together - Phase 2
Discovering Dengue Drugs - Together - Phase 2 (Type A)
Beta - The Clean Energy Project - Phase 2 is not available for your type of computer.

did you set this display up just so you could show no work available since those projects selected i can defiantly understand not seeing work.

if the project dose not have work after multiple requests BOINC (depending on version) adds time between connects it is normaly very short but i have seen on some projects getting a backing of 24 hours or longer do to no works..

This isn't the usual "bumping up" of the interval between requests when there is no work available.

It's not a big deal, just was showing that it wasn't only happening for Keith.

My guess is that it is somehow tied to the server message for the current linux-only beta. I'm running only windows machines here.

This is the first time that I can recall ever seeing a server message that beta work is not available.

To not land in a situation there clock on clients computer runs slightly faster than the server-clock, something that would mean client can always be "too resent", the deferral is multiplied with 1.01 before sent to client. So, with a server-config of 10 seconds, to the client the deferral is 11 seconds.


Quota exceeded deferrs to midnight + randomly upto 1 hour, so can be 24 hours. In cases of unsupported OS or too old BOINC-client it will AFAIK still give a 24-hour deferral. Also, if there's been 10 failed scheduler-requests in a row, the home web-page is tried re-downloaded, and if this also fails you'll get a 24-hour deferral. With older clients this last deferral was longer, AFAIK 1 week, possible even 2 weeks for really old clients...


There's 3 types of deferrals, the 11-second (or longer) asked by project, the random client-deferral that starts on 1 minute and increases upto 4 hours that happens if client can't connect to project, and is re-set to 1 minute after 10 connectsion. And, v6.6.xx and later clients introduced a new deferral, this you only sees by selecting project and looking on "preferences". (you'll also see it if you're running with <work_fetch_debug>).

This last deferral is per resource-type, so there's independent timers for cpu, ati and cuda. If a work-request for a specific resource is successful, the deferral-intervall is set to zero, and there's no deferral (if project doesn't ask for a deferral, like WCG does for GPU-work). If project did respond, but didn't give-out any work, the deferral-intervall starts at 1 minute, and for each following failing to get work it's doublet, upto a max of 24 hours. Also, for each failure, the client does a random backoff, between zero seconds and the current deferral-intervall.

It's re-set to zero each time client did get work for this resource-type, or user manually hits "update". And, since the reason for not getting work can be due project using max_wus_in_progress, each time a task does finish it's reset to zero again.



But back to the 1-hour deferral...
This will normally happen if project is shut-down for maintanance, or scheduling-server is up but there's a problem connecting to database and so on.

If project is completely out of work and client doesn't report any results, it's also possible to not hit the database at all, and give a 1-hour deferral...

There was also a 1-hour deferral for some other things, like not enough memory and not enough free disk space, but these deferrals was removed since this is per-wu, and the limits can be different for the next wu.
Among these wu-rules is also the "unsupported OS", that could give a 1-hour deferral, but this rule together with the rest should have been removed. So, most likely it's another instance with WCG running very old server-code...

Every time you don't get any work, but there was one (or more) beta-wu's available but you can't get these due to "wrong" platform, you'll be hit with the 1-hour deferral...

Hmm, so next time there's a windows-only beta, and anyone has selected projects that doesn't have work, the linux/mac-computers should get the 1-hour deferral.