Back Again (Aug 13 2009)

I was actually out the past couple of days. Family stuff, including an adventure where we had to tow our Prius almost 100 miles back to Oakland (it freaked out and lost power on I-5). It's in the shop now - luckily these newfangled cars store debugging information so they were able to locate the problem (flakey potentiometer causing erratic accelerator information, and as a failsafe the Prius cut its own power).

Anyway.. during the past couple of days Jeff and Bob handled the Tuesday outage, and Eric tackled a couple general network issues as well (the upload server got misconfigured somehow and was dropping excess connections, and then the assimilators were dead in the water for a while there, causing the queue to back up, the workunit disks to fill up, and finally the splitters to shut down - which is why we ran out of work to send out last night). All seems much better now, albeit jammed with traffic.

In better news we did finally get the first two data drives from Arecibo as recorded by the upgraded data recorder and new external drive docks under normal operations. So we're not going to run out of raw data after all, or at least not just yet. I'm copying those raw data files onto our local drives as I type.

- Matt

____________
-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

Thanks for the update Matt,

Claggy

I'm copying those raw data files onto our local drives as I type.

- Matt

So the pipes will be full for awhile and uploads will be jammed until we get all the AP WU out of the way. Looking at the server page it looks like 5-10 tapes so a blockage for around 24h or so. Thanks for the update Matt.
____________

I'm copying those raw data files onto our local drives as I type.

- Matt

any idea when these will go into crunch mode....

I'm copying those raw data files onto our local drives as I type.

Wot? No jobs?
Since about 06:30bst my PCs have been reporting "Message from server (Project has no jobs available)". Depending on which PC I look at this message is being sent every few minutes to about once an hour.
Are there no jobs on the servers, or is this a problem with a server?
Looking at the server status all looks well, but no jobs are forthcoming at my end of the string for the last 24 hours......
____________
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

I know that a bunch of short tasks passed through about two days ago all of my shorties have been validated. Although everything minus a few mb splitters are green on the Server Status I think a system is not working correctly somewhere. I expect this to be loose at Monday Berkeley time & Tuesday New Zealand time. I'm out of Seti tasks so I'm helping World Community Grid out.
____________

Live in NZ y not join Smile City?

I am runing only CPDN models, one long range and one medium range. All my other 5 BOINC projects are not sending any work. Einstein had a filesystem crash.
Tullio

My CPDN deadlines are December 2011 for hadcm3 and July 2010 for hadam3p. All my TDCF are less than 1 except for AQUA where it is 40. I was running all 5 projects up to a few days ago when I had hadcm3 already running, I then downloaded hadam3p, which should take 130 hours on my Linux box. I have completed 15% of hadcm3 and 25% of hadam3p.
Tullio

My CPDN deadlines are December 2011 for hadcm3 and July 2010 for hadam3p. All my TDCF are less than 1 except for AQUA where it is 40. I was running all 5 projects up to a few days ago when I had hadcm3 already running, I then downloaded hadam3p, which should take 130 hours on my Linux box. I have completed 15% of hadcm3 and 25% of hadam3p.
Tullio

I've noticed that issue with CPDN as well. As soon as I got my most recent CPDN task most of the projects refused to send work, despite the CPDN deadline 1 year in the future and the crunch time being under a week. I think something in the work fetch was changed and now it refuses to realize it has plenty of time to crunch the CPDN models. I'm sure this is a BOINC bug, likely a bug in the design and not the implementation. Reading some of the Wiki for it, I'm pretty sure the work fetch was designed to only work when projects are set with the same resource share and have work units that are similar in crunch time and deadlines that are about the same. I don't think the designers ever expected say LHC which 99% of the time won't have work, but might be set to crunch 99% by a user so he gets a work unit if one ever becomes available. This screws up the assumptions behind the work fetch and it simply misbehaves. I suspect the fetch algorithm is in two places doing a backoff. First when it tries to calculate how much time will be available, but assumes that the queue will be filled by other projects without this check being made before they fill the queue. Of course they do hence the second backoff. The second problem is the short/long term debt isn't being adjusted when a project doesn't send work because it thinks it won't finish in time.

Schedulers are a PITA and I think the work fetch needs a going over with a fine tooth comb and some limit testing. I think the improvements to it over time introduced subtle bugs.

There is also a bug when a user changes his work priorities, the short and long term debt isn't also changed, which means that work won't be crunched in the ratio the user just set.

There is a bug in 6.6.36 that can cause CPDN to take over completely. It is fixed in 6.6.38.