job: fix loss of ordering with restart jobs

Suppose that foo.service/start is a job waiting on other job bar.service/start
to finish. And then foo.service/restart is enqueued (not using
--ignore-dependencies).

Currently this makes foo.service start immediately, forgetting about the
ordering to bar.service.

The runnability check for JOB_RESTART jobs looks only at dependencies for
stopping. That's actually correct, because restart jobs should be treated the
same as stop jobs at first. The bug is that job_run_and_invalidate() does not
treat them exactly the same as stop jobs. unit_start() gets called without
checking for the runnability of the converted JOB_START job.

The fix is to simplify the switch in job_run_and_invalidate(). Handle
JOB_RESTART identically to JOB_STOP.
Also simplify the handling of JOB_TRY_RESTART - just convert it to JOB_RESTART
if the unit is active and let it fall through to the JOB_RESTART case.
Similarly for JOB_RELOAD_OR_START - have a fall through to JOB_START.

In job_finish_and_invalidate() it's not necessary to check for JOB_TRY_RESTART
with JOB_DONE, because JOB_TRY_RESTART jobs will have been converted to
JOB_RESTART already.

Speeding up the restart of services in "auto-restart" state still works as
before.

Improves: https://bugzilla.redhat.com/show_bug.cgi?id=753586
but it's still not perfect. With this fix the try-restart action will wait for
the restart to complete in the right order, but the optimal behaviour would be
to finish quickly (without disturbing the start job).
This commit is contained in:
Michal Schmidt 2012-03-28 00:42:27 +02:00
parent 6030831d5b
commit dd17d38879

View file

@ -387,14 +387,21 @@ int job_run_and_invalidate(Job *j) {
switch (j->type) {
case JOB_RELOAD_OR_START:
if (unit_active_state(j->unit) == UNIT_ACTIVE) {
j->type = JOB_RELOAD;
r = unit_reload(j->unit);
break;
}
j->type = JOB_START;
/* fall through */
case JOB_START:
r = unit_start(j->unit);
/* If this unit cannot be started, then simply
* wait */
/* If this unit cannot be started, then simply wait */
if (r == -EBADR)
r = 0;
break;
case JOB_VERIFY_ACTIVE: {
@ -408,11 +415,19 @@ int job_run_and_invalidate(Job *j) {
break;
}
case JOB_TRY_RESTART:
if (UNIT_IS_INACTIVE_OR_DEACTIVATING(unit_active_state(j->unit))) {
r = -ENOEXEC;
break;
}
j->type = JOB_RESTART;
/* fall through */
case JOB_STOP:
case JOB_RESTART:
r = unit_stop(j->unit);
/* If this unit cannot stopped, then simply
* wait. */
/* If this unit cannot stopped, then simply wait. */
if (r == -EBADR)
r = 0;
break;
@ -421,43 +436,6 @@ int job_run_and_invalidate(Job *j) {
r = unit_reload(j->unit);
break;
case JOB_RELOAD_OR_START:
if (unit_active_state(j->unit) == UNIT_ACTIVE) {
j->type = JOB_RELOAD;
r = unit_reload(j->unit);
} else {
j->type = JOB_START;
r = unit_start(j->unit);
if (r == -EBADR)
r = 0;
}
break;
case JOB_RESTART: {
UnitActiveState t = unit_active_state(j->unit);
if (t == UNIT_INACTIVE || t == UNIT_FAILED || t == UNIT_ACTIVATING) {
j->type = JOB_START;
r = unit_start(j->unit);
} else
r = unit_stop(j->unit);
break;
}
case JOB_TRY_RESTART: {
UnitActiveState t = unit_active_state(j->unit);
if (t == UNIT_INACTIVE || t == UNIT_FAILED || t == UNIT_DEACTIVATING)
r = -ENOEXEC;
else if (t == UNIT_ACTIVATING) {
j->type = JOB_START;
r = unit_start(j->unit);
} else {
j->type = JOB_RESTART;
r = unit_stop(j->unit);
}
break;
}
default:
assert_not_reached("Unknown job type");
}
@ -536,7 +514,7 @@ int job_finish_and_invalidate(Job *j, JobResult result) {
job_add_to_dbus_queue(j);
/* Patch restart jobs so that they become normal start jobs */
if (result == JOB_DONE && (j->type == JOB_RESTART || j->type == JOB_TRY_RESTART)) {
if (result == JOB_DONE && j->type == JOB_RESTART) {
log_debug("Converting job %s/%s -> %s/%s",
j->unit->id, job_type_to_string(j->type),