summary refs log tree commit homepage
diff options
context:
space:
mode:
authorEric Wong <e@80x24.org>2016-06-13 22:24:56 +0000
committerEric Wong <e@80x24.org>2016-06-13 22:24:56 +0000
commit40ac0012df419f6b7d3fae8ab86f7119abf877c2 (patch)
treeae2cbcbbc631b58b45d002c6d98b33f94db171c6
parent33c2f7794dc72b59390e6199fa0a6f41fb36eeb3 (diff)
By default, systemd kills every process in the control group
when stopping a service.  While it ought to be harmless to
signal workers, some Rack applications (and perhaps further
subprocesses) can misbehave when interrupted by a signal.
Ensure we only hit the master on graceful shutdown to avoid
tickling bugs in Rack apps.

This is the reason we switched to having the master send
"fake" signals for workers beginning with unicorn 4.8.0
back in 2013/2014.
-rw-r--r--examples/unicorn@.service5
1 files changed, 5 insertions, 0 deletions
diff --git a/examples/unicorn@.service b/examples/unicorn@.service
index 56aaec8..d95eb83 100644
--- a/examples/unicorn@.service
+++ b/examples/unicorn@.service
@@ -24,5 +24,10 @@ ExecReload = /bin/kill -HUP $MAINPID
 # adding a few seconds for scheduling differences:
 TimeoutStopSec = 62
 
+# Only kill the master process, it may be harmful to signal
+# workers via default "control-group" setting since some
+# Ruby extensions and applications misbehave on interrupts
+KillMode = process
+
 [Install]
 WantedBy = multi-user.target