Advanced search

Message boards : News : [wildlife] new runs up and going

Author Message
Travis Desell
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 16 Jan 12
Posts: 1809
Combined Credit: 21,510,416
DNA@Home: 293,563
SubsetSum@Home: 349,212
Wildlife@Home: 20,867,642
Wildlife@Home Watched: 212,926s
Wildlife@Home Events: 51
Climate Tweets: 21
Images Observed: 722

              
Message 7514 - Posted: 14 May 2018, 18:21:49 UTC

Hi All,

I've started up some new runs. These are using a different, larger, dataset, so let me know if you see any issues with them.

thanks!

Travis Desell
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 16 Jan 12
Posts: 1809
Combined Credit: 21,510,416
DNA@Home: 293,563
SubsetSum@Home: 349,212
Wildlife@Home: 20,867,642
Wildlife@Home Watched: 212,926s
Wildlife@Home Events: 51
Climate Tweets: 21
Images Observed: 722

              
Message 7518 - Posted: 17 May 2018, 18:43:33 UTC - in response to Message 7517.

This has caused the RAM requirement to bump to 1.5G, which in turn stops some of my clients from being able to download and run any tasks. These are 1gig cloud servers with 1gig of RAM, I have had to redirect some of them at another project.

17-May-2018 01:05:45 [Citizen Science Grid] Sending scheduler request: To fetch work. 17-May-2018 01:05:45 [Citizen Science Grid] Requesting new tasks for CPU 17-May-2018 01:05:48 [Citizen Science Grid] Scheduler request completed: got 0 new tasks 17-May-2018 01:05:48 [Citizen Science Grid] No tasks sent 17-May-2018 01:05:48 [Citizen Science Grid] EXACT Batch Norm With Scaled FMP CNN Trainer needs 1500.00 MB RAM but only 896.74 MB is available for use. 17-May-2018 01:33:09 [Citizen Science Grid] Sending scheduler request: To fetch work. 17-May-2018 01:33:09 [Citizen Science Grid] Requesting new tasks for CPU 17-May-2018 01:33:12 [Citizen Science Grid] Scheduler request completed: got 0 new tasks 17-May-2018 01:33:12 [Citizen Science Grid] No tasks sent 17-May-2018 01:33:12 [Citizen Science Grid] EXACT Batch Norm With Scaled FMP CNN Trainer needs 1500.00 MB RAM but only 896.74 MB is available for use. 17-May-2018 02:10:36 [Citizen Science Grid] Sending scheduler request: To fetch work. 17-May-2018 02:10:36 [Citizen Science Grid] Requesting new tasks for CPU 17-May-2018 02:11:07 [Citizen Science Grid] Scheduler request completed: got 0 new tasks 17-May-2018 02:11:07 [Citizen Science Grid] No tasks sent 17-May-2018 02:11:07 [Citizen Science Grid] EXACT Batch Norm With Scaled FMP CNN Trainer needs 1500.00 MB RAM but only 896.74 MB is available for use. 17-May-2018 02:55:35 [Citizen Science Grid] Sending scheduler request: To fetch work. 17-May-2018 02:55:35 [Citizen Science Grid] Requesting new tasks for CPU 17-May-2018 02:55:38 [Citizen Science Grid] Scheduler request completed: got 0 new tasks 17-May-2018 02:55:38 [Citizen Science Grid] No tasks sent 17-May-2018 02:55:38 [Citizen Science Grid] EXACT Batch Norm With Scaled FMP CNN Trainer needs 1500.00 MB RAM but only 896.74 MB is available for use.


After awhile (I'm assuming after someone else has downloaded these work units) I might get a new one, which follows the above log lines:

17-May-2018 05:32:48 [Citizen Science Grid] Sending scheduler request: To fetch work. 17-May-2018 05:32:48 [Citizen Science Grid] Requesting new tasks for CPU 17-May-2018 05:32:51 [Citizen Science Grid] Scheduler request completed: got 1 new tasks 17-May-2018 05:32:53 [Citizen Science Grid] Started download of exact_genome_1525431002_140_17717.txt 17-May-2018 05:32:55 [Citizen Science Grid] Finished download of exact_genome_1525431002_140_17717.txt 17-May-2018 05:32:55 [Citizen Science Grid] Starting task exact_genome_1525431002_140_17717_2


FYI, this is unfortunately going to decrease my contributions to CSG. As I'm using a group pool manager, the ability to micromanage which types of tasks I get from the project is unavailable to my instances.



Unfortunately, all the reviews I've had from recent publications say we need to move to larger datasets; so there's not anything I can do about the RAM requirements.

We had a couple issues with the server struggling to keep up with some of the previous workunits plus all the new work, and we're working on these. Work flow should at least be improving while we get these sorted out.


Post to thread

Message boards : News : [wildlife] new runs up and going