It seems to be about 3:30 am January 12 currently (utc), so the outage will be in one day and 18 hours from now, correct?

The outage will only last about 10.5 hours?

For the stats part, Kevin will be able to reconstruct the Friday stats, albeit with a 2.5 hour production shortfall, which then will be included in Saturday stats, or whenever the validators have been able to catch up on the scheduled 10.5 hour backlog, meaning the crunch is on to cram 26.5 hours of result checking into 16 hours. By implication, if the systems go live sometimes Saturday, the morning stats will be low too.

Now I'm wondering, will he have something in place to let clients back off a minimum amount of time to place the next connect attempt at sometimes after 08:00 UTC on Saturday? Watch those server messages in the client log and plz, do not try to hit the update button. You wont be served ;O)

Want to schedule the retrying of a reconnect and put the network to sleep automatically? You can.

1. In the activity menu set the *Network based on preferences*, if not already running like that:
2. In local Computing preferences, not web prefs, on the network usage tab tick the boxes next to Friday & Saturday and enter corrected for your local offset time to UTC the from / to times. For me with 1 hour ahead of UTC that's

00:00-22:30 (for Friday)
09:00-00:00 (for Saturday)

Notably, some client versions do not allow 24:00 as ending time. Enter 00:00

Those that normally run on website device profile preferences need to hit the clear button in the local prefs screen [on Saturday when services are up again] to remove the override file. If you don't, any later profile changes to networking and processing options made on the website will not take effect.

Thinking about this, and me plan to reconnect later I'll be entering:

00:00-22:00
14:00-00:00

(CET Times)

Then sometime before the next Friday will restore the old settings.

How to do this quickly when managing more than a few clients? Use BOINCTasks, the multi client manager. Runs on Windows, Linux and Mac.

--//--

Still confused. How can we get a WU downloaded, since the file system is up, if the scheduler is down. Doesn't a user have to request a WU via the scheduler before the file system sends the WU?

I am glad that uploads will still work, this means that I can just up my cache a little bit then lower it back before the outage. This will prevent my machine from making unnecessary communication to a down server.

OK. I got it [Lawrence too] what Kevin meant with the bolded part:

1. The download part that are the static files... the noted image files, the science application etc. These don't need the scheduler, so strictly speaking, if one joins a new device there will be coming down a bunch, but nothing related to a specific work unit.

2. The upload part, is what I assumed to be the data [that what is seen in the transfer tab of the client] This again does not require access to the central scheduler around which everything in a BOinC system pivots.

Net, work request wont be honored, Ready to Report wont clear (the step 2 part of the result reporting).

Tutti chiaro.

As for the shutdown, z000, that wont be until *Friday* evening for members on the European Continent. For those in North America it will be in the late afternoon, early evening. Note that UTC times are same as Greenwich Mean Time... where they put lime and milk in the tea [shudder].

--//--