Posts by Mike Cebula
1)
(Message 7225)
Posted 10 Jun 2021 by ![]() Post: One item on my Wish List is to limit the number of WUs downloaded to any individual computer. As an example, simply allow selecting 0 (one per thread), 1 day, 2 days or 3 days worth of work units. At this moment, and for several days now, there are over 2,500 WUs in progress. Browsing through several of them I see individual computers that still have over a hundred to process and have failed to meet the deadline on hundreds more. These have been reassigned and have extended the deadline beyond what was originally reasonable. The owner's of these WUs can only be frustrated by this turnaround. This is ridiculous. The second item on my Wish List is that a WU not be counted as an Error when the server cancels it. Exceeding the deadline or an incorrect computation is an error. I see almost all of my error WUs were started and then cancelled just minutes or a little over an hour into the work because the original computer finally uploaded its answer - beyond the deadline. If you miss the deadline, you should be cancelled immediately and the new computer should be able to do the entire job without fear of being cancelled and wasting its time. Comments? Mike Cebula |
2)
(Message 7176)
Posted 26 Mar 2021 by ![]() Post: I am not quite sure what some individuals are trying to prove, but I see where some computers have over a hundred (even over 300) tasks in progress on individual computers. Eventually, many of these seem to get the "Not started by deadline - canceled" message, and someone else picks up the slack. This does nothing for those of us waiting for the "Competed and validated" notation. Seems the algorithm for deciding limits per computer doesn't exist. All one has to do is set 'Computing Preferences' for a large number of days to get assigned a huge quantity of tasks and maybe have work if the host is down again for failures as has been the recent case. In any case, only the Administrator and staff can solve this. Anyone else care to comment on this? Mike |
3)
(Message 7150)
Posted 24 Feb 2021 by ![]() Post: Interesting. That directory on my machine doesn't contain any links to /var/lib/boinc-client. It only has two files: stderrgui.txt stdoutgui.txt Also, Asteroids isn't the only project active - at least when it is normally running. Also run Rosetta, Einstein and World Community Grid. None seem to have anything in the two files pointing elsewhere. Should also mention that I am running Linux Mint Mate 20.1 Mike |
4)
(Message 7147)
Posted 22 Feb 2021 by ![]() Post: Where, in Linux, can I find ALL the files associated with a specific project? So far I have identified the following: These are folders that contain easily identified files: var/lib/boinc-client/projects/...... var/lib/boinc-client/notices/...... var/lib/boinc-client/slots/...... The following are start of filenames associated with a project: var/lib/boinc-client/account...... var/lib/boinc-client/job_log...... var/lib/boinc-client/master...... var/lib/boinc-client/sched...... var/lib/boinc-client/statistics...... Thanks, Mike |
5)
(Message 7129)
Posted 6 Feb 2021 by ![]() Post: It looks to me like the university is treating asteroids@home as more of a hobby than a prestigious scientific effort. The cost for this 'used' server seems well within any type of budget the university may have for scientific education. Mike |
6)
(Message 7103)
Posted 21 Jan 2021 by ![]() Post: I have mine set for "no new tasks" with the idea being for all the completed tasks to be uploaded first before any new tasks are loaded. Seems like the cleanest way to not have a whole bunch of false starts and/or 'too late' work units. Maybe stay that way for a day or two until all the uploads have a chance. Mike |
7)
(Message 6967)
Posted 13 Jul 2020 by ![]() Post: What happens to the work units that have been sitting waiting to upload but are days past their original due date? Mike |
8)
(Message 6932)
Posted 11 Jul 2020 by ![]() Post: Number of users in the last 24 hours is down to 108. Should be getting someone's attention soon ... one would think. Mike |
9)
(Message 6930)
Posted 11 Jul 2020 by ![]() Post: Why would you abort them after they are finished? Mike |
10)
(Message 6904)
Posted 10 Jul 2020 by ![]() Post: I have a guess at what might happen with these in the current situation. Normally, if you are not done by the deadline, the work unit is assigned to someone else. However, if you finish before they do AND you get it uploaded, it will count and their work will be aborted by the server. However, if this problem of uploading persists and they finish the work, I believe the first one to finally get the work unit uploaded will get the credit. Of course, if you abort your finished work units, you will absolutely lose the work credit. No sense in aborting anything for now. Make sense? Mike |
11)
(Message 6900)
Posted 9 Jul 2020 by ![]() Post: All of Radim Vanco's tweets are in Czech. I picked the most recent one and replied in English that things were not 'stable' and he needs to fix this quickly and to look at the forums. Mike |
12)
(Message 6899)
Posted 9 Jul 2020 by ![]() Post: Not really fixed. I have a couple dozen work units trying to upload and the site says 'In progress'. Still a ways to go to get everything 'stable'. They will probably expire soon. Mike |
13)
(Message 6853)
Posted 8 Jul 2020 by ![]() Post: Already had some work units expire. One machine has 24 waiting for upload. Will reassign to Einstein@home and Rosetta@home until I see some progress with these administrators. Mike |
14)
(Message 6326)
Posted 2 Aug 2019 by ![]() Post: Host 621013 has an incompatible CPU: AuthenticAMD AMD Turion(tm) 64 X2 Mobile Technology TL-58 [Family 15 Model 104 Stepping 2], AuthenticAMD AMD Turion(tm) 64 X2 Mobile Technology TL-58 [Family 15 Model 104 Stepping 2] This is the reason given for not allowing the merge. Makes no sense. Mike |
15)
(Message 6325)
Posted 31 Jul 2019 by ![]() Post: I have updated one of my PCs and would really like to merge an older ID with the current one. Older ID is 621013 and it is now ID'd as 622745. The merge function doesn't work. The difference is one vs two processors (a repair) and am using an earlier Linux kernel version. The 621013 has a number of tasks 'in process' that will never be completed unless they are attached to 622745. This is actually the same machine. Is this possible? Mike |
16)
(Message 6009)
Posted 25 Oct 2018 by ![]() Post: What 'Sarge' said. Mike |