|
From: Jiri J. <jja...@re...> - 2014-09-23 09:44:28
|
From: Ondrej Moris <om...@re...>
On a virtual machines it might happen due to potential caching of
/var/log/audit mounted on tmpfs that /sbin/init is called too many
times during disk_full halt and single tests. This caused very long
testing time and due to imjournal rate-limit it moreover caused
imjournal to drop messages. With this commit we kill audisp-remote
(causing /sbin/init to be triggered) as soon as possible (ie. after
first execution of /sbin/init) from /sbin/init itself.
Signed-off-by: Ondrej Moris <om...@re...>
---
audit-test/audit-remote/tests/audisp-remote_functions.bash | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/audit-test/audit-remote/tests/audisp-remote_functions.bash b/audit-test/audit-remote/tests/audisp-remote_functions.bash
index f64c420..dcbca67 100644
--- a/audit-test/audit-remote/tests/audisp-remote_functions.bash
+++ b/audit-test/audit-remote/tests/audisp-remote_functions.bash
@@ -541,6 +541,10 @@ pre_halt() {
#!/bin/bash -x
if [[ \$1 == $mask_runlevel ]]; then
logger "audisp-remote_testing: $$ runlevel=\$1"
+
+ # Since /sbin/init might be executed many times, we should
+ # kill audisp-remote to avoid unpleasant chain-reaction
+ killall -q audisp-remote
else
exec /sbin/init.audisp-remote_testing "\$@"
fi
--
1.8.3.1
|