to make a checkpoint

Questions and Answers : Getting started : to make a checkpoint

To post messages, you must log in.

AuthorMessage
akin

Send message
Joined: 7 Apr 07
Posts: 1
Credit: 44,540
RAC: 0
Message 39454 - Posted: 15 Apr 2007, 22:05:02 UTC

Hello

Boinc is great but my english is not :)

i can't find a way to save my work when i turn off computer when i restart my computer the whole process start from begining...

is there any kind of to make check point for rosetta cos electrical conditions is not well in here and sometimes i have 2-3 blackout in a day

please help. Thanx
ID: 39454 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 39474 - Posted: 16 Apr 2007, 2:36:16 UTC

Unstable power would be a definate challenge. Some types of work have checkpoints within a model. Others only checkpoint at the end of each model. The Project Team is working to add additional checkpoints in their next release in a couple of weeks. This should help preserve more of your work.

There really is not anything on your side that you can change... unless of course you get a battery backup or something that keeps your machine running :)

All you can do is set your General Preference to leave applications in memory while preempted. And avoid powering down your machine.

Also, the main way to preserve work is to progress further in to working on a task. This will be true even when additional checkpointing is added. Since you have unstable power, I would suggest you not run many BOINC projects on a single machine. Having multiple projects means halting one to work on another, and each time you halt a project, you may have halted it at a point where work has not been preserved. Since you have 3 machines, I would suggest you divide your time between projects by setting up one project on each machine, rather then having several projects on each machine.
Rosetta Moderator: Mod.Sense
ID: 39474 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile shanen
Avatar

Send message
Joined: 16 Apr 14
Posts: 195
Credit: 12,662,308
RAC: 0
Message 78881 - Posted: 5 Oct 2015, 12:36:55 UTC

Ancient problem, NEVER fixed.

Hey, here's a stupid suggestion. A manual checkpoint option.

And here's the smart suggestion. Shutting down the computer normally should make sure there are checkpoints on all running work units.

Just a perennial annoyance, but it's a major part of why I dropped the previous project I had been supporting after earning something over a million "Work done" units. Currently at 1.4 million on rosetta@home, but getting increasingly annoyed with work units that have no checkpoints for an hour or so at a time.

For security and performance reasons I prefer not sleeping the computers, in case you were going to repeat that suggestion...
ID: 78881 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 78887 - Posted: 7 Oct 2015, 3:37:53 UTC

There is no single fix to bring more frequent checkpointing. Each type of protein analysis has it's own structure and limitations for when checkpoints are taken. Taking a checkpoint is not like hitting save on an otherwise static spreadsheet. R@h tasks literally have thousands of things going on as they run. The processing hits points where things quiet down, especially at the end of a model. When such points are reached where a checkpoint can be taken and accurately restarted, checkpoints are taken. As specific approaches to analyze a protein are used more or future plans to use a given approach are made, additional checkpoints are added at some of the more difficult points in the processing of a model; or other times enhancements are made to reduce the runtime required per model (and checkpoints are always taken when a given model is completed).

Are there specific patterns in the WU names where you are seeing the long times between checkpoints?
Rosetta Moderator: Mod.Sense
ID: 78887 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Questions and Answers : Getting started : to make a checkpoint



©2024 University of Washington
https://www.bakerlab.org