dm-devel.redhat.com archive mirror
 help / color / mirror / Atom feed
From: Benjamin Marzinski <bmarzins@redhat.com>
To: mwilck@suse.com
Cc: lixiaokeng@huawei.com, dm-devel@redhat.com,
	Chongyun Wu <wu.chongyun@h3c.com>
Subject: Re: [dm-devel] [PATCH 07/35] multipathd: improve delayed reconfigure
Date: Wed, 15 Sep 2021 18:00:46 -0500	[thread overview]
Message-ID: <20210915230045.GS3087@octiron.msp.redhat.com> (raw)
In-Reply-To: <20210910114120.13665-8-mwilck@suse.com>

On Fri, Sep 10, 2021 at 01:40:52PM +0200, mwilck@suse.com wrote:
> From: Martin Wilck <mwilck@suse.com>
> 
> When a reconfigure operation is requested, either by the admin
> or by some condition multipathd encounters, the current code
> attempts to set DAEMON_CONFIGURE state and gives up after a second
> if it doesn't succeed. Apart from shutdown, this happens only
> if multipathd is either already reconfiguring, or busy in the
> path checker loop.
> 
> This patch modifies the logic as follows: rather than waiting,
> we set a flag that requests a reconfigure operation asap, i.e.
> when the current operation is finished and the status switched
> to DAEMON_IDLE. In this case, multipathd will not switch to IDLE
> but start another reconfigure cycle.
> 
> This assumes that if a reconfigure is requested while one is already
> running, the admin has made some (additional) changes and wants
> multipathd to pull them in. As we can't be sure that the currently
> running reconfigure has seen the configuration changes, we need
> to start over again.
> 
> A positive side effect is less waiting in clients and multipathd.
> 
> After this change, the only caller of set_config_state() is
> checkerloop(). Waking up every second just to see that DAEMON_RUNNING
> couldn't be set makes no sense. Therefore set_config_state() is
> changed to wait "forever", or until shutdown is requested. Unless
> multipathd completely hangs, the wait will terminate sooner or
> later.
> 
> Signed-off-by: Martin Wilck <mwilck@suse.com>
> ---
>  multipathd/cli_handlers.c | 10 +----
>  multipathd/main.c         | 92 +++++++++++++++++++++++++++++----------
>  multipathd/main.h         |  3 +-
>  3 files changed, 71 insertions(+), 34 deletions(-)
> 
> diff --git a/multipathd/cli_handlers.c b/multipathd/cli_handlers.c
> index 6d3a0ae..44f76ee 100644
> --- a/multipathd/cli_handlers.c
> +++ b/multipathd/cli_handlers.c
> @@ -1076,17 +1076,9 @@ cli_switch_group(void * v, char ** reply, int * len, void * data)
>  int
>  cli_reconfigure(void * v, char ** reply, int * len, void * data)
>  {
> -	int rc;
> -
>  	condlog(2, "reconfigure (operator)");
>  
> -	rc = set_config_state(DAEMON_CONFIGURE);
> -	if (rc == ETIMEDOUT) {
> -		condlog(2, "timeout starting reconfiguration");
> -		return 1;
> -	} else if (rc == EINVAL)
> -		/* daemon shutting down */
> -		return 1;
> +	schedule_reconfigure();
>  	return 0;
>  }
>  
> diff --git a/multipathd/main.c b/multipathd/main.c
> index 67160b9..5fb6989 100644
> --- a/multipathd/main.c
> +++ b/multipathd/main.c
> @@ -221,6 +221,10 @@ static void do_sd_notify(enum daemon_status old_state,
>  	} else if (new_state == DAEMON_CONFIGURE && startup_done)
>  		sd_notify(0, "RELOADING=1");
>  }
> +#else
> +static void do_sd_notify(__attribute__((unused)) enum daemon_status old_state,
> +			 __attribute__((unused)) enum daemon_status new_state)
> +{}
>  #endif
>  
>  static void config_cleanup(__attribute__((unused)) void *arg)
> @@ -266,19 +270,38 @@ enum daemon_status wait_for_state_change_if(enum daemon_status oldstate,
>  	return st;
>  }
>  
> +/* Don't access this variable without holding config_lock */
> +static bool reconfigure_pending;
> +
>  /* must be called with config_lock held */
>  static void __post_config_state(enum daemon_status state)
>  {
>  	if (state != running_state && running_state != DAEMON_SHUTDOWN) {
> -#ifdef USE_SYSTEMD
>  		enum daemon_status old_state = running_state;
> -#endif
>  
> +		/*
> +		 * Handle a pending reconfigure request.
> +		 * DAEMON_IDLE is set from child() after reconfigure(),
> +		 * or from checkerloop() after completing checkers.
> +		 * In either case, child() will see DAEMON_CONFIGURE
> +		 * again and start another reconfigure cycle.
> +		 */
> +		if (reconfigure_pending && state == DAEMON_IDLE &&
> +		    (old_state == DAEMON_CONFIGURE ||
> +		     old_state == DAEMON_RUNNING)) {
> +			/*
> +			 * notify systemd of transient idle state, lest systemd
> +			 * thinks the reload lasts forever.
> +			 */
> +			do_sd_notify(old_state, DAEMON_IDLE);
> +			old_state = DAEMON_IDLE;
> +			state = DAEMON_CONFIGURE;
> +		}
> +		if (reconfigure_pending && state == DAEMON_CONFIGURE)
> +			reconfigure_pending = false;
>  		running_state = state;
>  		pthread_cond_broadcast(&config_cond);
> -#ifdef USE_SYSTEMD
>  		do_sd_notify(old_state, state);
> -#endif
>  	}
>  }
>  
> @@ -290,24 +313,48 @@ void post_config_state(enum daemon_status state)
>  	pthread_cleanup_pop(1);
>  }
>  
> -int set_config_state(enum daemon_status state)
> +void schedule_reconfigure(void)
> +{
> +	pthread_mutex_lock(&config_lock);
> +	pthread_cleanup_push(config_cleanup, NULL);
> +	switch (running_state)
> +	{
> +	case DAEMON_SHUTDOWN:
> +		break;
> +	case DAEMON_IDLE:
> +		__post_config_state(DAEMON_CONFIGURE);
> +		break;
> +	case DAEMON_CONFIGURE:
> +	case DAEMON_RUNNING:
> +		reconfigure_pending = true;
> +		break;
> +	default:
> +		break;
> +	}
> +	pthread_cleanup_pop(1);
> +}
> +
> +enum daemon_status set_config_state(enum daemon_status state)
>  {
>  	int rc = 0;
> +	enum daemon_status st;
>  
>  	pthread_cleanup_push(config_cleanup, NULL);
>  	pthread_mutex_lock(&config_lock);
> -	if (running_state != state) {
>  
> -		if (running_state == DAEMON_SHUTDOWN)
> -			rc = EINVAL;
> -		else
> -			rc = __wait_for_state_change(
> -				running_state != DAEMON_IDLE, 1000);
> -		if (!rc)
> -			__post_config_state(state);
> +	while (rc == 0 &&
> +	       running_state != state &&
> +	       running_state != DAEMON_SHUTDOWN &&
> +	       running_state != DAEMON_IDLE) {
> +		rc = pthread_cond_wait(&config_cond, &config_lock);
>  	}
> +
> +	if (rc == 0 && running_state == DAEMON_IDLE && state != DAEMON_IDLE)
> +		__post_config_state(state);
> +	st = running_state;
> +
>  	pthread_cleanup_pop(1);
> -	return rc;
> +	return st;
>  }
>  
>  struct config *get_multipath_config(void)
> @@ -734,7 +781,7 @@ ev_add_map (char * dev, const char * alias, struct vectors * vecs)
>  			if (delayed_reconfig &&
>  			    !need_to_delay_reconfig(vecs)) {
>  				condlog(2, "reconfigure (delayed)");
> -				set_config_state(DAEMON_CONFIGURE);
> +				schedule_reconfigure();
>  				return 0;
>  			}
>  		}
> @@ -1845,7 +1892,7 @@ missing_uev_wait_tick(struct vectors *vecs)
>  	if (timed_out && delayed_reconfig &&
>  	    !need_to_delay_reconfig(vecs)) {
>  		condlog(2, "reconfigure (delayed)");
> -		set_config_state(DAEMON_CONFIGURE);
> +		schedule_reconfigure();
>  	}
>  }
>  
> @@ -2484,6 +2531,10 @@ checkerloop (void *ap)
>  		int num_paths = 0, strict_timing, rc = 0;
>  		unsigned int ticks = 0;
>  
> +		if (set_config_state(DAEMON_RUNNING) != DAEMON_RUNNING)
> +			/* daemon shutdown */
> +			break;
> +
>  		get_monotonic_time(&start_time);
>  		if (start_time.tv_sec && last_time.tv_sec) {
>  			timespecsub(&start_time, &last_time, &diff_time);
> @@ -2499,13 +2550,6 @@ checkerloop (void *ap)
>  		if (use_watchdog)
>  			sd_notify(0, "WATCHDOG=1");
>  #endif
> -		rc = set_config_state(DAEMON_RUNNING);
> -		if (rc == ETIMEDOUT) {
> -			condlog(4, "timeout waiting for DAEMON_IDLE");
> -			continue;
> -		} else if (rc == EINVAL)
> -			/* daemon shutdown */
> -			break;
>  
>  		pthread_cleanup_push(cleanup_lock, &vecs->lock);
>  		lock(&vecs->lock);
> @@ -2833,7 +2877,7 @@ handle_signals(bool nonfatal)
>  		return;
>  	if (reconfig_sig) {
>  		condlog(2, "reconfigure (signal)");
> -		set_config_state(DAEMON_CONFIGURE);
> +		schedule_reconfigure();
>  	}
>  	if (log_reset_sig) {
>  		condlog(2, "reset log (signal)");
> diff --git a/multipathd/main.h b/multipathd/main.h
> index bc1f938..23ce919 100644
> --- a/multipathd/main.h
> +++ b/multipathd/main.h
> @@ -37,6 +37,7 @@ void exit_daemon(void);
>  const char * daemon_status(void);
>  enum daemon_status wait_for_state_change_if(enum daemon_status oldstate,
>  					    unsigned long ms);
> +void schedule_reconfigure(void);
>  int need_to_delay_reconfig (struct vectors *);
>  int reconfigure (struct vectors *);
>  int ev_add_path (struct path *, struct vectors *, int);
> @@ -44,7 +45,7 @@ int ev_remove_path (struct path *, struct vectors *, int);
>  int ev_add_map (char *, const char *, struct vectors *);
>  int ev_remove_map (char *, char *, int, struct vectors *);
>  int flush_map(struct multipath *, struct vectors *, int);
> -int set_config_state(enum daemon_status);
> +enum daemon_status set_config_state(enum daemon_status);

Can't we just remove set_config_state from main.h, and make it static?
Other than that, everything looks fine.

-Ben

>  void * mpath_alloc_prin_response(int prin_sa);
>  int prin_do_scsi_ioctl(char *, int rq_servact, struct prin_resp * resp,
>  		       int noisy);
> -- 
> 2.33.0

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel


  reply	other threads:[~2021-09-15 23:03 UTC|newest]

Thread overview: 89+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-09-10 11:40 [dm-devel] [PATCH 00/35] multipathd: uxlsnr overhaul mwilck
2021-09-10 11:40 ` [dm-devel] [PATCH 01/35] libmultipath: add timespeccmp() utility function mwilck
2021-09-15 22:07   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 02/35] libmultipath: add trylock() helper mwilck
2021-09-15 22:07   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 03/35] libmultipath: add optional wakeup functionality to lock.c mwilck
2021-09-15 22:13   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 04/35] libmultipath: print: add __snprint_config() mwilck
2021-09-15 22:14   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 05/35] libmultipath: improve cleanup of uevent queues on exit mwilck
2021-09-15 22:20   ` Benjamin Marzinski
2021-09-16  7:10     ` Martin Wilck
2021-09-16 14:26       ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 06/35] multipathd: fix systemd notification when stopping while reloading mwilck
2021-09-15 22:55   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 07/35] multipathd: improve delayed reconfigure mwilck
2021-09-15 23:00   ` Benjamin Marzinski [this message]
2021-09-16  7:16     ` Martin Wilck
2021-09-10 11:40 ` [dm-devel] [PATCH 08/35] multipathd: cli.h: formatting improvements mwilck
2021-09-15 23:01   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 09/35] multipathd: cli_del_map: fix reply for delayed action mwilck
2021-09-15 23:40   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 10/35] multipathd: add prototype for cli_handler functions mwilck
2021-09-15 23:53   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 11/35] multipathd: make all cli_handlers static mwilck
2021-09-15 23:53   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 12/35] multipathd: add and set cli_handlers in a single step mwilck
2021-09-16  0:01   ` Benjamin Marzinski
2021-09-16  7:22     ` Martin Wilck
2021-11-12 21:45     ` Martin Wilck
2021-09-10 11:40 ` [dm-devel] [PATCH 13/35] multipathd: cli.c: use ESRCH for "command not found" mwilck
2021-09-16  0:02   ` Benjamin Marzinski
2021-09-10 11:40 ` [dm-devel] [PATCH 14/35] multipathd: add "force_reconfigure" option mwilck
2021-09-16  0:13   ` Benjamin Marzinski
2021-09-16  7:34     ` Martin Wilck
2021-09-16 14:32       ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 15/35] multipathd: uxlsnr: avoid stalled clients during reconfigure mwilck
2021-09-16  2:17   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 16/35] multipathd: uxlsnr: handle client HUP mwilck
2021-09-16  2:17   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 17/35] multipathd: uxlsnr: use symbolic values for pollfd indices mwilck
2021-09-16  2:18   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 18/35] multipathd: uxlsnr: avoid using fd -1 in ppoll() mwilck
2021-09-16  2:18   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 19/35] multipathd: uxlsnr: data structure for stateful client connection mwilck
2021-09-16  2:19   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 20/35] multipathd: move uxsock_trigger() to uxlsnr.c mwilck
2021-09-16  2:19   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 21/35] multipathd: move parse_cmd() " mwilck
2021-09-16  2:19   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 22/35] multipathd: uxlsnr: remove check_timeout() mwilck
2021-09-16  2:21   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 23/35] multipathd: uxlsnr: move client handling to separate function mwilck
2021-09-16  2:21   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 24/35] multipathd: uxlsnr: use main poll loop for receiving mwilck
2021-09-16  2:22   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 25/35] multipathd: use strbuf in cli_handler functions mwilck
2021-09-16  2:23   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 26/35] multipathd: uxlsnr: check root on connection startup mwilck
2021-09-16  2:23   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 27/35] multipathd: uxlsnr: pass struct client to uxsock_trigger() and parse_cmd() mwilck
2021-09-16  2:28   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 28/35] multipathd: uxlsnr: move handler execution to separate function mwilck
2021-09-16  2:28   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 29/35] multipathd: uxlsnr: use parser to determine non-root commands mwilck
2021-09-16  2:29   ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 30/35] multipathd: uxlsnr: merge uxsock_trigger() into state machine mwilck
2021-09-16  3:32   ` Benjamin Marzinski
2021-09-16  8:02     ` Martin Wilck
2021-11-12 22:07     ` Martin Wilck
2021-09-10 11:41 ` [dm-devel] [PATCH 31/35] multipathd: uxlsnr: add idle notification mwilck
2021-09-16  4:14   ` Benjamin Marzinski
2021-09-16  8:54     ` Martin Wilck
2021-09-16 15:06       ` Benjamin Marzinski
2021-09-16 15:54         ` Martin Wilck
2021-09-16 16:10           ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 32/35] multipathd: uxlsnr: add timeout handling mwilck
2021-09-16  4:17   ` Benjamin Marzinski
2021-09-16  8:58     ` Martin Wilck
2021-09-16 15:08       ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 33/35] multipathd: uxlsnr: use poll loop for sending, too mwilck
2021-09-16  4:22   ` Benjamin Marzinski
2021-09-16  9:33     ` Martin Wilck
2021-09-16 15:26       ` Benjamin Marzinski
2021-09-10 11:41 ` [dm-devel] [PATCH 34/35] multipathd: uxlsnr: drop client_lock mwilck
2021-09-16  4:24   ` Benjamin Marzinski
2021-09-16  9:34     ` Martin Wilck
2021-09-10 11:41 ` [dm-devel] [PATCH 35/35] multipathd: uxclt: allow client mode for non-root, too mwilck
2021-09-16  4:24   ` Benjamin Marzinski

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210915230045.GS3087@octiron.msp.redhat.com \
    --to=bmarzins@redhat.com \
    --cc=dm-devel@redhat.com \
    --cc=lixiaokeng@huawei.com \
    --cc=mwilck@suse.com \
    --cc=wu.chongyun@h3c.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).