Bug #3279
Load spiked in event.c for scheduler
| Status: | Closed | Start date: | 08/15/2012 | |
|---|---|---|---|---|
| Priority: | Urgent | Due date: | ||
| Assignee: | % Done: | 0% |
||
| Category: | Scheduler | |||
| Target version: | 2.1.0 | |||
| Rank: | Tester: |
Description
The use of g_async_queue_timed_pop() is incorrect for versions of glib-2.0 < 2.34 in src/scheduler/agent/event.c. The timeout value should be current time plus the timeout duration. Example:
g_get_current_time(&timeout);
g_time_val_add(&timeout, 1000000);
if((ret = g_async_queue_timed_pop(vl->queue, &timeout)) == NULL)
return ret;
History
Updated by Mary Laser 9 months ago
- Assignee set to Alex Norton
- Target version set to 2.1.0
not sure if you'll get to this before your new job starts. If not, please reassign.
Updated by Bob Gobeille 9 months ago
- Priority changed from Normal to Urgent
- Estimate set to 2
- IterNum set to 7
We might need to release a 2.0.1 version of the scheduler for this one.
Updated by Bob Gobeille 9 months ago
from IRC:
[14:29] <bobg> bapper, you referred to glib 2.34 in your bug. I thought the latest (unstable) was 2.32.4
[14:30] <bapper> bobg: oops! 2.32 is what it should be.
[14:30] <bapper> bobg: #if GLIB_MAJOR_VERSION >= 2 && GLIB_MINOR_VERSION >= 32
[14:30] <bapper> that's the statement in the code.
Updated by Alex Norton 9 months ago
fixed with svn 6146.
I added the line suggested, Bob or Marry need to confirm that the scheduler is no longer running at %100 cpu on systems that are using a glib version lower than 2.32
Updated by Alex Norton 9 months ago
- Status changed from New to Resolved
Updated by Bob Gobeille 9 months ago
- Status changed from Resolved to Closed
tested on older system that was previously exhibiting the problem