diff options
| author | Tom Ritter <tom@ritter.vg> | 2017-11-11 23:25:15 -0500 |
|---|---|---|
| committer | Tom Ritter <tom@ritter.vg> | 2017-11-11 23:25:15 -0500 |
| commit | ee3eea047c10d0032efc5a8b98d9bbdb54fe0a3a (patch) | |
| tree | f7208e5c3e1107280153e6c90933f300c6113441 /samplejobs | |
| parent | 6371a93650e74d6d4518890230eeeb5ffa7b660e (diff) | |
Fix executions that stomped on each other
Every hour I would fire off two executions: the minute execution and the hour execution. Each would read the statefile on start.
Whichever execution took longer would overwrite the state file with the data that was present for the other's execution upon start.
This would manifest as extra notifications if the starts aligned - because the 'I already sent a notification' member update
wasn't being persisted.
This fixes it by only performing one execution at any given time, but knowing to run all the relevant jobs for that type of execution.
AKA When I run the hourly job I also run the minute jobs.
Another problem related to this one is logged in #1 - minute executions that take longer than a minute will also cause this type of error.
Diffstat (limited to 'samplejobs')
0 files changed, 0 insertions, 0 deletions