Boinc and Java
19 May 2010 3:38:54 UTC
Topic 194935
(moderation:
The other day, a browser window was open on my desktop and had java running. The java icon was down in my system tray and Boinc was running also. About every hour, Boinc said
17-May-2010 21:08:37 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
17-May-2010 21:08:37 [Einstein@Home] If this happens repeatedly you may need to reset the project.
17-May-2010 21:08:38 [Einstein@Home] Restarting task h1_0451.00_S5R4__137_S5GC1a_1 using einstein_S5GC1 version 302
17-May-2010 22:06:11 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
17-May-2010 22:06:11 [Einstein@Home] If this happens repeatedly you may need to reset the project.
17-May-2010 22:06:23 [Einstein@Home] Restarting task h1_0451.00_S5R4__137_S5GC1a_1 using einstein_S5GC1 version 302
18-May-2010 00:05:04 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
18-May-2010 00:05:04 [Einstein@Home] If this happens repeatedly you may need to reset the project.
18-May-2010 00:05:07 [Einstein@Home] Restarting task h1_0451.00_S5R4__137_S5GC1a_1 using einstein_S5GC1 version 302
18-May-2010 01:02:06 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
18-May-2010 01:02:06 [Einstein@Home] If this happens repeatedly you may need to reset the project.
18-May-2010 01:02:06 [Einstein@Home] Restarting task h1_0451.00_S5R4__137_S5GC1a_1 using einstein_S5GC1 version 302
18-May-2010 02:01:01 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
18-May-2010 02:01:01 [Einstein@Home] If this happens repeatedly you may need to reset the project.
18-May-2010 02:01:02 [Einstein@Home] Restarting task h1_0451.00_S5R4__137_S5GC1a_1 using einstein_S5GC1 version 302
18-May-2010 02:59:57 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
18-May-2010 02:59:57 [Einstein@Home] If this happens repeatedly you may need to reset the project.
18-May-2010 02:59:58 [Einstein@Home] Restarting task h1_0451.00_S5R4__137_S5GC1a_1 using einstein_S5GC1 version 302
18-May-2010 03:57:47 [Einstein@Home] Task h1_0451.00_S5R4__137_S5GC1a_1 exited with zero status but no 'finished' file
18-May-2010 03:57:47 [Einstein@Home] If this happens repeatedly you may need to reset the project.
18-May-2010 03:57:58 [Einstein@Home]
Once the browser window was closed and java ended, the message never returned. My task validated but I was wondering why this would've happened or if it was something else going on? I can reproduce it at any time. The task id was 174916453.
Boinc and Java
)
Apart from resource contention I can't think of any evident reason. The JVM can be hoggy sometimes ...
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: Apart from resource
)
A bit far-fetched perhaps, but tinkering with the system clock can produce that effect too.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
RE: RE: Apart from
)
Hi Gundolf:
Not at all far-fetched, I think this is indeed the problem. See this excerpt from one of caveman's logs:
Note the negative time shift.
CU
HBE
RE: Not at all far-fetched,
)
Or, more importantly note the heartbeat message ...
No heartbeat, the Science Application suicides ...
This is a long-standing issue with BOINC, it is, and has been, a critical rated issue in the Trac database as bug #6 (I think) since like forever ... you can also get it from running other projects that have large files to un-zip (or zip) during their task set-up or ends...
I did a whole proof series on Drug Discovery and Hydrogen causing this problem (also IBERCIVIS) last year or so ... The JVM for one reason or another was taking so much CPU that BOINC could not get a word in edge-wise and the science applications lost contact with the mothership and thought that BOINC had died ... so they killed themselves as rogue processes ... JVM stopped being a hog and BOINC noticed the science apps were all dead and restarted them ...
My wife likes to play Text
)
My wife likes to play Text Twist on Yahoo and will finish a round and leave the application open till she can get back to it later in the day. This leaves the JVM on all day and sometimes all night in the task manager the Einstein app is using 94 to 99 percent of the cpu while the JVM sits there idle. Even with it like this running overnight with no user interaction on the computer, this seems to happen every hour. Since the task validated, am I correct to say that even with these periodic restarts, everything is fine with it running this way?
RE: My wife likes to play
)
I doubt that the JAVA VM is to blame here. The problem seems to be that once per hour, your PC (either a utility application or a setting in Windows) will synchronize the real-time clock with some external clock over the internet. It happens that BOINC isn't coping with this time drift very well.
Is this a serious problem? No. As this is happening only max 1 time per hour, the tasks will finally finish ok. Only if this type of error happens more frequently, BOINC will think it's a realy problem and will abort the tasks.
CU
HBE
The limit is 100 restarts.
)
The limit is 100 restarts. Under that and you are fine...
I was vaguely thinking of
)
I was vaguely thinking of system timers as a resource, but I don't know enough about the JVM internals to be sure of how it uses that. And that would be ( real ) machine dependent too .....
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Since there is not a problem
)
Since there is not a problem with the science being done, I will just continue to crunch away. I was just curious as to how or why it was happening and if any harm may arise from it. Probably just a quirk in my computer setup. I turned off my router and isolated the computer from being online today to see if it was connecting and adjusting the system clock, but the same thing continued to happen roughly every hour. Oh well. Thank you all for the helpful information.