rake resque:pool isn't respecting TERM_CHILD?

[adambedford]

I think I'm seeing an issue where rake resque:pool doesn't honor the TERM_CHILD=1 flag. If I run resque-pool, it does work as expected.

Here is the output from rake resque:pool:

~/app $ TERM_CHILD=1 bundle exec rake resque:pool
resque-pool-worker[integrations][34943]: Starting worker abedford-mbpr.local:34943:*
resque-pool-worker[integrations][34944]: Starting worker abedford-mbpr.local:34944:*
resque-pool-manager[integrations][34935]: started manager
resque-pool-worker[integrations][34945]: Starting worker abedford-mbpr.local:34945:*
resque-pool-manager[integrations][34935]: Pool contains worker PIDs: [34943, 34944, 34945, 34946, 34947]
resque-pool-worker[integrations][34946]: Starting worker abedford-mbpr.local:34946:*
resque-pool-worker[integrations][34947]: Starting worker abedford-mbpr.local:34947:*
^Cresque-pool-manager[integrations][34935]: INT: immediate shutdown (graceful worker shutdown)
resque-pool-manager[integrations][34935]: Sending TERM to all workers
resque-pool-manager[integrations][34935]: manager finished

Here is the output from resque-pool:

~/app $ TERM_CHILD=1 bundle exec resque-pool
resque-pool-manager[integrations][34858]: Resque Pool running in development environment
resque-pool-manager[34858]: TERM_CHILD enabled, so will user 'term-graceful-and-wait' behaviour
resque-pool-worker[integrations][34866]: Starting worker abedford-mbpr.local:34866:*
resque-pool-worker[integrations][34867]: Starting worker abedford-mbpr.local:34867:*
resque-pool-worker[integrations][34868]: Starting worker abedford-mbpr.local:34868:*
resque-pool-manager[integrations][34858]: started manager
resque-pool-manager[integrations][34858]: Pool contains worker PIDs: [34866, 34867, 34868, 34869, 34870]
resque-pool-worker[integrations][34870]: Starting worker abedford-mbpr.local:34870:*
resque-pool-worker[integrations][34869]: Starting worker abedford-mbpr.local:34869:*
^Cresque-pool-manager[integrations][34858]: INT: immediate shutdown (graceful worker shutdown)
resque-pool-manager[integrations][34858]: Sending TERM to all workers
resque-pool-manager[integrations][34858]: manager finished

I notice this line: resque-pool-manager[34858]: TERM_CHILD enabled, so will user 'term-graceful-and-wait' behaviour indicating that term child is enabled. Does the lack of this in the rake resque:pool mean what I think it does - that TERM_CHILD isn't respected?

[codyrobbins]

I ran into this same issue—running rake resque:pool seems to ignore TERM_CHILD entirely and the only way to have it take effect is to run resque-pool.