Browse code
use nproc/2 workers for large ops job
Commit 1ce19ab76d67a89b04f907f1d292d013a3b699e0 dropped API_WORKERS from
nproc/2 to nproc/4 and also started using API_WORKERS for the number of
conductor workers, so in gate runs that dropped conductor workers from 8
to 2. We're now seeing instance build timeouts in the large ops job.
This change goes back to nproc/2 for the large ops job (VIRT_DRIVER=='fake').
Closes-Bug: #1500615
Change-Id: Ie6ef855fce0a99c930d479b7459c15b69e8de499
Matt Riedemann authored on 2015/09/29 06:46:27Showing 1 changed files
| ... | ... |
@@ -650,7 +650,12 @@ SERVICE_PROTOCOL=${SERVICE_PROTOCOL:-http}
|
| 650 | 650 |
# the memory used where there are a large number of CPUs present |
| 651 | 651 |
# (the default number of workers for many services is the number of CPUs) |
| 652 | 652 |
# Also sets the minimum number of workers to 2. |
| 653 |
-API_WORKERS=${API_WORKERS:=$(( ($(nproc)/4)<2 ? 2 : ($(nproc)/4) ))}
|
|
| 653 |
+if [[ "$VIRT_DRIVER" = 'fake' ]]; then |
|
| 654 |
+ # we need more workers for the large ops job |
|
| 655 |
+ API_WORKERS=${API_WORKERS:=$(( ($(nproc)/2)<2 ? 2 : ($(nproc)/2) ))}
|
|
| 656 |
+else |
|
| 657 |
+ API_WORKERS=${API_WORKERS:=$(( ($(nproc)/4)<2 ? 2 : ($(nproc)/4) ))}
|
|
| 658 |
+fi |
|
| 654 | 659 |
|
| 655 | 660 |
# Service startup timeout |
| 656 | 661 |
SERVICE_TIMEOUT=${SERVICE_TIMEOUT:-60}
|