Doesn't make any sense at all.

What we NEED is the current project to work without constant new hitches!

It seems that working down the purging of old valid WUs is almost done, there could be some hope that things normalize (in a rational way) and hopefully stay that way.

I have the feeling somehow that that competition that is supposed to start on Monday might throw a wrench in the wheels of trying to move forward consistently.
But for now, it seems there is again a steady supply of MCM1 WUs and let's see what really happens with OPNG WUs on Monday or if the vultures descend from the skies and ruin it once again for all the rest...


Ralf

There is only a very small amount of WUs being issued. It's been like this for days.

Hopefully the problems at Krembill will soon be rectified.

I wonder how many of the folks not seeing a regular flow of work are falling foul of the unfortunate tendency of the BOINC client to assume that failure to get work implies a possible problem at the server end, so the next request is delayed, and if that also fails to get work the next delay is longer (and so on, up to a point where the delay is many hours!)

Given that [at present] there seem to be frequent periods of over an hour where no new work is being released, that could result in the client only happening to check in when there's no work so the client continues to assume it should delay...

I only mention this because I use a script to make sure that my clients ask for work at least 3 times an hour and I only see a fall-off in work supplied when no new work is created for several hours... The work is out there most of the time, it can just be quite hard to find it :-)

Cheers - Al.

P.S. I run short queues and have low per-application limits set in my device profiles, so if there's no new work for [say] 6 hours my cores will start to starve. Before the great assimilation backlog, this was only an issue if there was a genuine server problem such as "feeder not running" :-)

I have 18-19 machines but had (2) Linux boxes (Debian v12) in particular that had run dry as far as WCG was concerned.

One hadn't seen a WU since 03/09 when I began turning-on the Backup Projects (but leaving WCG open). Sadly I noticed that machine had rolled-off the Device List on the website. I hope it returns after fixing it via the below...

--

To fix these two I allowed the Backup Projects to complete & placed on "No new tasks"; then removed WCG Project from the BOINC Client; rebooted the systems; changed the "Store at least x-days" to 0.2 under Computing Preferences; and told it to "Run Always" under the Activity menu to prevent any non-WCG site from messing with my preferences.

I then added WCG and waited as it re-downloaded all the *.PNG graphics like a new Project does.

After that, on the next Communication it picked-up 2-3 WU's. I increased the x-number of days back and so far (knock on wood), those (2) machines are running.

--

I had one Windows machine that aborted several WU's saying they were overdue, but still had 4-days to complete. That one stumped me, but I rebooted the system and it went back to working normally (though it still sent the aborted tasks back), so I assume that was a BOINC Client corruption issue.

--

Seems like each weekend I spend 1-2 hours managing these WU's on all my machines which is troubling since BOINC should be "set it and forget it"

73's...

On two of my twelve computers, tasks are absent, while on the remaining ten, each equipped with sixteen cores, numerous cores lie idle