mirror of
https://github.com/troglobit/finit.git
synced 2026-11-20 19:41:51 +00:00
service: start restart always at two seconds between restarts
A crashing service with restart always waited five seconds before
every restart. It should wait two seconds for the first five, like a
service with the default restart-max:
service foo {
restart = "always"
command = "foo"
}
The back-off starts after half of restart-max attempts. For restart
always restart-max is -1, so half of it was zero and the back-off
began with the first restart. Use the default restart-max instead.
Signed-off-by: Joachim Wiberg <troglobit@gmail.com>
This commit is contained in:
5 files changed
+67
-5
No files matched your search
@@ -188,6 +188,8 @@ All relevant changes are documented in this file.
|
||||
- Fix a command whose name starts like a service option, e.g. `logger` or
|
||||
`pidof`. It was taken as the `log` or `pid` option and the next argument
|
||||
ran as the command, which also broke the `logger` example in finit.conf(5)
|
||||
- Fix `restart = "always"` skipping the initial two second delay between
|
||||
restarts, every restart waited five seconds
|
||||
- Reject a .conf file that declares the same block title twice. The title is
|
||||
the service identity and libConfuse merges two sections sharing one, so the
|
||||
file loaded as a single service holding a mix of both declarations,
|
||||
|
||||
@@ -176,8 +176,9 @@ following settings are available:
|
||||
* `restart-sec = SEC` -- number of seconds before Finit tries to
|
||||
restart a crashing service. The default is 2 seconds for the first
|
||||
half of `restart-max` attempts, then a back-off to 5 seconds -- with
|
||||
the default `restart-max` that is the first five retries. The
|
||||
greater of this configured value and the back-off is used
|
||||
the default `restart-max`, and with `restart = "always"`, that is the
|
||||
first five retries. The greater of this configured value and the
|
||||
back-off is used
|
||||
* `restart = "always"` -- no upper limit on the number of times Finit
|
||||
tries to restart a crashing service
|
||||
* `restart = "never"` -- do not restart on failures. `false` is
|
||||
|
||||
+4
-3
@@ -3267,7 +3267,7 @@ static void service_cleanup_script(svc_t *svc)
|
||||
static void service_retry(svc_t *svc)
|
||||
{
|
||||
char *restart_cnt = (char *)&svc->restart_cnt;
|
||||
int timeout;
|
||||
int timeout, half;
|
||||
|
||||
service_timeout_cancel(svc);
|
||||
|
||||
@@ -3326,8 +3326,9 @@ static void service_retry(svc_t *svc)
|
||||
dbg("%s crashed, trying to start it again, attempt %d", svc_ident(svc, NULL, 0), *restart_cnt);
|
||||
if ((*restart_cnt) == 1)
|
||||
svc->restart_saved = svc->restart_tmo;
|
||||
/* Wait 2s for the first 5 respawns, then back off to 5s */
|
||||
timeout = ((*restart_cnt) <= (svc->restart_max / 2)) ? 2000 : 5000;
|
||||
/* Wait 2s for the first half of the attempts, then back off to 5s */
|
||||
half = (svc->restart_max == -1 ? SVC_RESPAWN_MAX : svc->restart_max) / 2;
|
||||
timeout = *restart_cnt <= half ? 2000 : 5000;
|
||||
/* If a longer timeout was specified in the conf, use that instead. */
|
||||
svc->restart_tmo = max(svc->restart_tmo, timeout);
|
||||
if (WIFEXITED(svc->status))
|
||||
|
||||
@@ -65,6 +65,7 @@ EXTRA_DIST += process-depends.sh
|
||||
EXTRA_DIST += rclocal.sh
|
||||
EXTRA_DIST += ready-serv.sh
|
||||
EXTRA_DIST += remain-task-control.sh
|
||||
EXTRA_DIST += restart-backoff.sh
|
||||
EXTRA_DIST += restart-self.sh
|
||||
EXTRA_DIST += runlevel.sh
|
||||
EXTRA_DIST += run-restart-forever.sh
|
||||
@@ -139,6 +140,7 @@ TESTS += process-depends.sh
|
||||
TESTS += rclocal.sh
|
||||
TESTS += ready-serv.sh
|
||||
TESTS += remain-task-control.sh
|
||||
TESTS += restart-backoff.sh
|
||||
TESTS += restart-self.sh
|
||||
TESTS += runlevel.sh
|
||||
TESTS += sysvparts.sh
|
||||
|
||||
Executable
+56
@@ -0,0 +1,56 @@
|
||||
#!/bin/sh
|
||||
# Verify the delay between restarts of a crashing service. It logs
|
||||
# the uptime, in centiseconds, of each start and then exits.
|
||||
set -eu
|
||||
|
||||
TEST_DIR=$(dirname "$0")
|
||||
|
||||
STARTS=/tmp/starts
|
||||
|
||||
# Wait for start $2, then check the delay since start $1, in sec
|
||||
assert_delay()
|
||||
{
|
||||
retry "texec sh -c 'test \$(wc -l < $STARTS) -ge $2'" 150 0.1
|
||||
# shellcheck disable=SC2016 # $1 is awk's field reference, not the shell's
|
||||
delay=$(texec awk -v a="$1" -v b="$2" 'NR == a { t = $1 } NR == b { print $1 - t }' $STARTS)
|
||||
assert "Delay between start $1 and $2 is $3-$4 sec" "$delay" -ge "$(($3 * 100))"
|
||||
assert "Delay between start $1 and $2 is $3-$4 sec" "$delay" -le "$(($4 * 100))"
|
||||
}
|
||||
|
||||
# Replace the service, settings in $1, and let it start crashing
|
||||
crash()
|
||||
{
|
||||
run "rm -f $FINIT_CONF"
|
||||
run "initctl reload"
|
||||
run "rm -f $STARTS"
|
||||
run "cat > $FINIT_CONF" <<CONF
|
||||
service crash {
|
||||
$1
|
||||
command = "crash.sh"
|
||||
}
|
||||
CONF
|
||||
run "initctl reload"
|
||||
}
|
||||
|
||||
test_teardown()
|
||||
{
|
||||
say "Running test teardown."
|
||||
run "rm -f $FINIT_CONF $STARTS /bin/crash.sh"
|
||||
}
|
||||
|
||||
# shellcheck source=/dev/null
|
||||
. "$TEST_DIR/lib/setup.sh"
|
||||
|
||||
cat > "$SYSROOT/bin/crash.sh" <<SCRIPT
|
||||
#!/bin/sh
|
||||
tr -d . < /proc/uptime | cut -d' ' -f1 >> $STARTS
|
||||
exit 1
|
||||
SCRIPT
|
||||
chmod +x "$SYSROOT/bin/crash.sh"
|
||||
|
||||
sep
|
||||
say 'restart always, at once the first time, then every other second'
|
||||
crash 'restart = "always"'
|
||||
assert_delay 1 2 0 1
|
||||
assert_delay 2 3 2 3
|
||||
assert_delay 3 4 2 3
|
||||
Reference in new issue
Block a user