summaryrefslogtreecommitdiffstats
path: root/monitor.c
diff options
context:
space:
mode:
authorDan Williams <dan.j.williams@intel.com>2010-05-14 17:42:49 -0700
committerDan Williams <dan.j.williams@intel.com>2010-05-14 17:42:49 -0700
commit484240d8a3facde992009efd81bfa4cc0c79287d (patch)
treee8caef63af97a57817ab331256528f0836737ea3 /monitor.c
parent63b4aae33ebf00d443378daf313622630f2336c0 (diff)
downloadmdadm-484240d8a3facde992009efd81bfa4cc0c79287d.tar.gz
mdadm-484240d8a3facde992009efd81bfa4cc0c79287d.tar.xz
mdadm-484240d8a3facde992009efd81bfa4cc0c79287d.zip
mdmon: periodically checkpoint recovery
The kernel updates and notifies md/sync_completed when it is time to take a checkpoint. When this occurs (at 1/16 array size intervals) write 'idle' to md/sync_action to have the current recovery position updated in recovery_start and resync_start. Requires the metadata handler to reset ->last_checkpoint when it has determined that recovery has ended. Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Diffstat (limited to 'monitor.c')
-rw-r--r--monitor.c33
1 files changed, 33 insertions, 0 deletions
diff --git a/monitor.c b/monitor.c
index e43e545..12f8d3e 100644
--- a/monitor.c
+++ b/monitor.c
@@ -80,6 +80,24 @@ static unsigned long long read_resync_start(int fd)
return strtoull(buf, NULL, 10);
}
+static unsigned long long read_sync_completed(int fd)
+{
+ unsigned long long val;
+ char buf[50];
+ int n;
+ char *ep;
+
+ n = read_attr(buf, 50, fd);
+
+ if (n <= 0)
+ return 0;
+ buf[n] = 0;
+ val = strtoull(buf, &ep, 0);
+ if (ep == buf || (*ep != 0 && *ep != '\n' && *ep != ' '))
+ return 0;
+ return val;
+}
+
static enum array_state read_state(int fd)
{
char buf[20];
@@ -195,6 +213,7 @@ static void signal_manager(void)
static int read_and_act(struct active_array *a)
{
+ unsigned long long sync_completed;
int check_degraded = 0;
int deactivate = 0;
struct mdinfo *mdi;
@@ -206,6 +225,7 @@ static int read_and_act(struct active_array *a)
a->curr_state = read_state(a->info.state_fd);
a->curr_action = read_action(a->action_fd);
a->info.resync_start = read_resync_start(a->resync_start_fd);
+ sync_completed = read_sync_completed(a->sync_completed_fd);
for (mdi = a->info.devs; mdi ; mdi = mdi->next) {
mdi->next_state = 0;
if (mdi->state_fd >= 0) {
@@ -307,6 +327,18 @@ static int read_and_act(struct active_array *a)
}
}
+ /* Check for recovery checkpoint notifications. We need to be a
+ * minimum distance away from the last checkpoint to prevent
+ * over checkpointing. Note reshape checkpointing is not
+ * handled here.
+ */
+ if (sync_completed > a->last_checkpoint &&
+ sync_completed - a->last_checkpoint > a->info.component_size >> 4 &&
+ a->curr_action > reshape && a->next_action == bad_action) {
+ a->last_checkpoint = sync_completed;
+ a->next_action = idle;
+ }
+
a->container->ss->sync_metadata(a->container);
dprintf("%s(%d): state:%s action:%s next(", __func__, a->info.container_member,
array_states[a->curr_state], sync_actions[a->curr_action]);
@@ -461,6 +493,7 @@ static int wait_and_act(struct supertype *container, int nowait)
add_fd(&rfds, &maxfd, a->info.state_fd);
add_fd(&rfds, &maxfd, a->action_fd);
+ add_fd(&rfds, &maxfd, a->sync_completed_fd);
for (mdi = a->info.devs ; mdi ; mdi = mdi->next)
add_fd(&rfds, &maxfd, mdi->state_fd);