same here in Belgium

Interesting. There's nothing new in Known Issues. About two hours ago, I was able to download new work fine. Updates process fine as well. Guess someone fixed it quick

I've been getting and sending work units all day without a problem. Just sent another to test it. No problems here.

Sorry - that was my fault.

Earlier this week we had to change the way that work is sent out. There is a certain user out there who added over 250 intel-macs to the grid. This user was processing about half of the work being done by intel-macs. Becuase we only send results for a certain workunit to the same type of machine (due to floating point round differences between platforms which makes validation of results nearly impossible otherwise) and becuase we were only allowing one result per workunit to be sent to a user, this meant that all of the other intel-macs were not able to process the extra two copies of each workunit. The backlog grew and other platforms started to have problems getting work.

So we had to temporarily disable this setting (which people noticed earlier in the week) until we could change the code so that it worked a 'device' level instead of a 'user' level (no computer will get multiple copies of work for the same workunit - however a user might). This resolved the problem which we deployed on Wednesday.

However, I noticed that the load on the database was much higher following this change and we needed to add a new index. That was done at the time that Johnny Cool noticed the problem. Previously I have been able to add indexes without impacting users - however it appears that with our higher user base we will need to make these changes during scheduled changes.

It took about 50 minutes to add the index so problems would have been experienced during that period. I apologize for the problems that you experienced.

Kevin

growth is also always a good problem