search

USDA-ARS-NWRC / smrf

SMRF was designed to increase the flexibility of taking measured weather data, or atmospheric models, and distributing the data across a watershed.
Other
12 stars 4 forks source link

Wind thread variables clean up #183

Closed scotthavens closed 4 years ago

scotthavens commented 4 years ago

Trying to clean up the thread variables for the wind variables to address #180

scotthavens commented 4 years ago

It's still happening...

jomey commented 4 years ago

It's still happening...

Was that locally? The error was from Coverall for the checks.

scotthavens commented 4 years ago

It's still happening...

Was that locally? The error was from Coverall for the checks.

It happened on the first run for 3.6, I restarted the job and then it failed on coveralls. This one has failed

scotthavens commented 4 years ago

The latest run failed again but it did not fail on the thread variables. So it appears that the thread variables are being set properly. This may point to the failure coming from the output or cleaner thread.

scotthavens commented 4 years ago

It's seem to have been fixed. I have ran the latest tests 3 times without failure (except for coveralls which is allowed to fail now). While the output queues were cleaned up, I think the actual fix was increasing the timeout to 5 seconds... I don't thing the Travis VM's are all that fast so the queues were timing out while SMRF was running.

jomey commented 4 years ago

It's seem to have been fixed. I have ran the latest tests 3 times without failure (except for coveralls which is allowed to fail now). While the output queues were cleaned up, I think the actual fix was increasing the timeout to 5 seconds... I don't thing the Travis VM's are all that fast so the queues were timing out while SMRF was running.

The timeout sounds very plausible. CI environments are prone to fluctuating test execution times depending on their load. On one other project I work on, the test times for the (very big) suite fluctuates between 70 and 180 minutes.

scotthavens commented 4 years ago

It's seem to have been fixed. I have ran the latest tests 3 times without failure (except for coveralls which is allowed to fail now). While the output queues were cleaned up, I think the actual fix was increasing the timeout to 5 seconds... I don't thing the Travis VM's are all that fast so the queues were timing out while SMRF was running.

The timeout sounds very plausible. CI environments are prone to fluctuating test execution times depending on their load. On one other project I work on, the test times for the (very big) suite fluctuates between 70 and 180 minutes.

Made some good improvements even it the execution time was the problem!