Major Outage Notice: Grid Offline 3:00 UTC Jan 22 - 3:00 UTC Jan 23

[OldChap]

I thought that everybody may not be reading the Server News thread so I just wanted to Headline this while there is still time to get everybody's cache filled!!

The title says it all.

[StitchExperimen]

from WCG

https://secure.worldcommunitygrid.or...d?thread=34590

In order to resolve an ongoing performance issue impacting the grid, we need to halt all access to the grid database starting at 3:00 UTC on Jan 22 through 3:00 UTC on Jan 23. During this time your computer will not be able to fetch new work to process and it will not be able to report complete work back to our servers. Additionally, in order to keep our filesystem from filling up, we will need to prevent files from being uploaded to our servers during this time as well. However, during this time the website and forums should remain available.

The technical details about why we need to take this outage are available here: http://www.worldcommunitygrid.org/fo...ad?post=408639

We apologize for the need to take this extended outage and we appreciate your patience as we work through this issue.

[Snow Crash]

Bump - your caches

[Sparky]

I'm gonna run out in a few short hours. Forgot to increase my cache. Crud.

[yojimbo197]

Thanks for the warning, Old Chap. I managed to increase my caches on my regular crunchers. It will be interesting to see how much of a pileup is present after a day.

[1200W]

I just got my 7950 in the machine and now this..OH gawd Why oh why! Anyways, i am ready to get some wu-s!

[hendermd]

Posts from Knreed

System is available again but experiencing massive load. It will take 20-30 minutes to stabilize.

[Jan 22, 2013 8:53:29 PM]

Due to an error, all preference updates were not correctly sent between databases. If you made a change to your preferences in the past 24 hours, please go to the preference page and re-save your settings again. That will cause them to be sent again.

[Jan 22, 2013 9:11:50 PM]

[OldChap]

I think I will wait until tomorrow to upload. I only used about 1/3rd of my cache

[wplachy]

Or you could suspend new tasks and then allow your completed WUs to upload. That give your wingmen a shot at validation

[OldChap]

All of the volume uploads will be finished in less than an hour from now ...then, for me, it is already tomorrow. I am on utc time.

My wingmen will get their share in line with the original estimate for how long this was going to take. I don't suppose that it is any more than 6000 results anyway.

I might even get a different flavour of pie for a day

[Johnmark]

All of the volume uploads will be finished in less than an hour from now ...then, for me, it is already tomorrow. I am on utc time.

My wingmen will get their share in line with the original estimate for how long this was going to take. I don't suppose that it is any more than 6000 results anyway.

I might even get a different flavour of pie for a day

OC, How did you prevent your completed wu's from uploading once the server was running again? Just curious

[OldChap]

In BOINC manager: Activity> Network Activity Suspended.

If I am near the rigs and I become aware of server issues which prevent upload I tend to do this to save a few cycles each time the system tries to connect.

First rig is now normal again and the others are uploading now too

EDIT: I have mine set to upload 20 at a time.... does anybody go higher than this??

[bluestang]

OC, how/where do you set that?

[Movieman]

Now that was a massive dump.....ahhhhhhh..
I can just imagine what happened when SAM and Hypernova uploaded..Can you see a server sitting there and starting to scream as it's circuits heat up to molten silicon!

[OldChap]

OC, how/where do you set that?

<cc_config>
<options>
<!--report results immediately, only use when project is running low, otherwise it stressed the servers (0 is default, 1 is immediate)-->
<report_results_immediately>0</report_results_immediately>
<!--Make sure this is larger than the per project number (8 by default)-->
<max_file_xfers>21</max_file_xfers>
<!-- Set this to the number of transfers you would like to allow at the same time (2 by default) -->
max_file_xfers_per_project>20</max_file_xfers_per_project>
</options>
<log_flags>
</log_flags>
</cc_config>

Note the two elements to this. In green is the maximum concurrent downloads on the rig. In red is the maximum per project

[bluestang]

Got it, both set at 20. Thanks, doing much faster now!

[StitchExperimen]

Is there a way if I was to make a workstation with dual processors to run 4 7950s? I once tried to make Boinc 2 and then a regular Boinc and it wouldn't let me have 2 programs.

[OldChap]

I would hope it would do it automatically. I have noticed that on some motherboards the slots are connected to particular cpu's so it may be that you need to put the cards into a specific slot.

Ultimately there is a thread here that shows how to run two instances of boinc. http://www.xtremesystems.org/forums/...pp_info-method

[Movieman]

Is there a way if I was to make a workstation with dual processors to run 4 7950s? I once tried to make Boinc 2 and then a regular Boinc and it wouldn't let me have 2 programs.

Just use one BOINC..PM me any details on board,chips,etc

[StitchExperimen]

Thanks OldChap