Message boards :
News :
Workplan
Message board moderation
| Author | Message |
|---|---|
 Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184  |
This thread will be used to provide information on the status of issues/improvements. Strikethrough will be used to show items done (check for issues), bold for in progress and italics for things on the to-do list. |
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
At some point, all the thousands of "ghosts" need to be cleared out of the user accounts. |
ivan Send message Joined: 20 Jan 15 Posts: 1129 Credit: 7,876,173 RAC: 266  |
At some point, all the thousands of "ghosts" need to be cleared out of the user accounts. I think they'll disappear when the time limit is reached. I believe it was one week for most, so just a few days to go... 
|
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
Tasks end after first run |
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
Will keep it as it is for now as things are working well and it will speed up the iterations if we need a new VM to update a configuration parameter. |
|
Send message Joined: 16 Aug 15 Posts: 966 Credit: 1,211,816 RAC: 0 |
Maybe it is really better to keep it this way, as there is no time saving involved having a 2nd run versus a new task. |
|
Send message Joined: 13 Feb 15 Posts: 1180 Credit: 815,336 RAC: 238  |
Will keep it as it is for now as things are working well and it will speed up the iterations if we need a new VM to update a configuration parameter. It's running fine now, only a bit more overhead when BOINC gets a new task and have to create a new VM and that new VM have to load data that else w(c)ould have been used for 4 runs (24 hours). One point if you'll keep this single run 'for ever'. The max duration of the last 100 BOINC-tasks is 26.37 hours. Do you plan to create a new CMS_2016_Month_Date.xml to reduce BOINC job_duration to let's say 12 or 18 hours. |
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
I am open to suggestions. The value for 24 hour tasks came from test4theory and as far as I understand a compromise between the overhead of restarting a VM and the desire for volunteers to see credit/task progress. We can change the length of the task and also the length of the the run. So instead of having 4 runs in 24 hours we could have 1 run that executes 16 jobs instead of 4. What task length would you suggest works best and is there any advantage of having multiple runs or should we just increase the number of jobs that a run does? |
|
Send message Joined: 13 Feb 15 Posts: 1180 Credit: 815,336 RAC: 238 |
There is another 'unknown' factor. The scientist decides how many events should be processed in 1 job. I prefer jobs with a duration of about 1 hour on an average computer. We don't have to decide how many jobs should be done in 1 glidein run. That depends on the computer speed. When the one and only glidein has run for 12 hours, it could send the shutdown (kill) to vboxwrapper after the last job has finished and just for sake BOINC's job_duration set to 18 hours for the reason, that BOINC is not obeying the shutdown. |
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
|
Laurence Send message Joined: 12 Sep 14 Posts: 1064 Credit: 328,405 RAC: 184 |
|
©2024 CERN