Anonymous | Login | Signup for a new account | 2024-09-01 02:43 EDT | ![]() |
My View | View Issues | Change Log | Roadmap |
View Issue Details [ Jump to Notes ] | [ Issue History ] [ Print ] | ||||||||
ID | Project | Category | View Status | Date Submitted | Last Update | ||||
0000414 | JANA | Bug | public | 2014-08-14 22:10 | 2014-08-14 22:13 | ||||
Reporter | davidl | ||||||||
Assigned To | davidl | ||||||||
Priority | normal | Severity | minor | Reproducibility | always | ||||
Status | resolved | Resolution | fixed | ||||||
Platform | OS | OS Version | |||||||
Summary | 0000414: Program exists if hi-res timer runs out | ||||||||
Description | Aristeidis reported this: For some especially long jobs (ones running for more than 1 day) the program would suddenly die with a message about a timer running out. This turns out to be due signals being sent when any of the 3 hi-res timers we use runs out. The default behavior for uncaught signals of these types is to terminate the program. The timers are already set for 1M seconds, but apparently, with multiple threads, this counts quicker than the wall clock, (Otherwise, they would run for over 270 hours before timing out.) To fix this, alarm handlers need to be installed for SIGALRM, SIGPROF, and SIGVTARLM signals. | ||||||||
Tags | No tags attached. | ||||||||
Attached Files | |||||||||
![]() |
|
(0000594) davidl (administrator) 2014-08-14 22:13 |
This fix was committed in revision 2075 and will appear in JANA 0.7.2 |
![]() |
|||
Date Modified | Username | Field | Change |
2014-08-14 22:10 | davidl | New Issue | |
2014-08-14 22:13 | davidl | Note Added: 0000594 | |
2014-08-14 22:13 | davidl | Status | new => resolved |
2014-08-14 22:13 | davidl | Resolution | open => fixed |
2014-08-14 22:13 | davidl | Assigned To | => davidl |
Copyright © 2000 - 2024 MantisBT Team |