[ovs-dev] [PATCH 2/2] daemon: Throttle max respawning rate.
Ben Pfaff
blp at nicira.com
Wed May 12 17:25:54 UTC 2010
On Tue, May 11, 2010 at 05:26:46PM -0700, Jesse Gross wrote:
> On Tue, May 11, 2010 at 11:15 AM, Ben Pfaff <blp at nicira.com> wrote:
> >
> > + /* Throttle restarts to no more than once every 10
> > seconds. */
> > + if (time(NULL) < last_restart + 10) {
> > + VLOG_WARN("%s, waiting until 10 seconds since last "
> > + "restart", status_msg);
> > + while (time(NULL) < last_restart + 10) {
> > + sleep(1);
> > + }
>
>
> Why do we sleep in a loop here rather than computing the amount of time
> once?
Mostly laziness. I figured we needed a loop in case sleep() wakes up
with EINTR. But sleeping for the remaining time is probably a better
approach.
Here's a revised patch. Do you like it better?
--8<--------------------------cut here-------------------------->8--
>From a3aa59700fc0f43dc4dd0d3acfd45392b0d8463d Mon Sep 17 00:00:00 2001
From: Ben Pfaff <blp at nicira.com>
Date: Wed, 12 May 2010 10:02:23 -0700
Subject: [PATCH] daemon: Throttle max respawning rate.
If a monitored daemon dies quickly at startup, the system can waste a lot
of CPU time continually restarting it. This commit prevents a given
daemon from restarting more than once every 10 seconds.
---
lib/daemon.c | 18 +++++++++++++++++-
1 files changed, 17 insertions(+), 1 deletions(-)
diff --git a/lib/daemon.c b/lib/daemon.c
index 292be14..78f88a0 100644
--- a/lib/daemon.c
+++ b/lib/daemon.c
@@ -323,14 +323,15 @@ should_restart(int status)
static void
monitor_daemon(pid_t daemon_pid)
{
- /* XXX Should limit the rate at which we restart the daemon. */
/* XXX Should log daemon's stderr output at startup time. */
const char *saved_program_name;
+ time_t last_restart;
char *status_msg;
saved_program_name = program_name;
program_name = xasprintf("monitor(%s)", program_name);
status_msg = xstrdup("healthy");
+ last_restart = TIME_MIN;
for (;;) {
int retval;
int status;
@@ -365,6 +366,21 @@ monitor_daemon(pid_t daemon_pid)
}
}
+ /* Throttle restarts to no more than once every 10 seconds. */
+ if (time(NULL) < last_restart + 10) {
+ VLOG_WARN("%s, waiting until 10 seconds since last "
+ "restart", status_msg);
+ for (;;) {
+ time_t now = time(NULL);
+ time_t wakeup = last_restart + 10;
+ if (now >= wakeup) {
+ break;
+ }
+ sleep(wakeup - now);
+ }
+ }
+ last_restart = time(NULL);
+
VLOG_ERR("%s, restarting", status_msg);
daemon_pid = fork_and_wait_for_startup(&daemonize_fd);
if (!daemon_pid) {
--
1.7.1
More information about the dev
mailing list