[Cluster-devel] cluster/cmirror-kernel/src dm-cmirror-client.c ...

[jbrassow@sourceware.org <jbrassow@sourceware.org>]

CVSROOT:	/cvs/cluster
Module name:	cluster
Branch: 	RHEL4
Changes by:	jbrassow at sourceware.org	2006-12-07 18:58:32

Modified files:
	cmirror-kernel/src: dm-cmirror-client.c dm-cmirror-server.c 

Log message:
	Get rid of a number of unnecessary messages, which spray to the console
	during errors and cause the mirror reconfiguration to take a long time.
	(Even been seen to cause machines to be fenced if the load is too great.)

Patches:
http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/cmirror-kernel/src/dm-cmirror-client.c.diff?cvsroot=cluster&only_with_tag=RHEL4&r1=1.1.2.32&r2=1.1.2.33
http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/cmirror-kernel/src/dm-cmirror-server.c.diff?cvsroot=cluster&only_with_tag=RHEL4&r1=1.1.2.17&r2=1.1.2.18

--- cluster/cmirror-kernel/src/Attic/dm-cmirror-client.c	2006/12/05 17:49:08	1.1.2.32
+++ cluster/cmirror-kernel/src/Attic/dm-cmirror-client.c	2006/12/07 18:58:32	1.1.2.33
@@ -905,9 +905,8 @@
 	spin_lock(&region_state_lock);
 	list_for_each_entry_safe(rs, tmp_rs, &clear_region_list, rs_list){
 		if(lc == rs->rs_lc && region == rs->rs_region){
-#ifdef DEBUG
-			DMINFO("Mark pre-empting clear of region %Lu", region);
-#endif
+			DMDEBUG("Mark pre-empting clear (%Lu/%s)",
+				region, lc->uuid + (strlen(lc->uuid) - 8));
 			list_del_init(&rs->rs_list);
 			list_add(&rs->rs_list, &marked_region_list);
 			clear_region_count--;
@@ -1023,7 +1022,7 @@
 	** clearing without ever marking..................................... */
 
 	if(!rs_new){
-		DMERR("Unable to allocate region_state for mark.");
+		DMERR("Unable to allocate region_state for clear.");
 		BUG();
 	}
 
@@ -1058,9 +1057,6 @@
 	while(consult_server(lc, region, LRT_COMPLETE_RESYNC_WORK, &success_tmp)){
 		DMWARN("unable to notify server of completed resync work");
 	}
-	if (!success)
-		DMERR("Attempting to revert sync status of region #%llu", region);
-
 	return;
 }
 
@@ -1069,11 +1065,7 @@
 	int i;
 	region_t rtn;
 	struct log_c *lc = (struct log_c *) log->context;
-/* take out optimization
-	if(atomic_read(&lc->in_sync) == 1){
-		return lc->region_count;
-	}
-*/
+
 	/* Try to get sync count up to five times */
 	for (i = 0; i < 5 && consult_server(lc, 0, LRT_GET_SYNC_COUNT, &rtn); i++);
 	if(i >= 5){
--- cluster/cmirror-kernel/src/Attic/dm-cmirror-server.c	2006/09/05 17:50:11	1.1.2.17
+++ cluster/cmirror-kernel/src/Attic/dm-cmirror-server.c	2006/12/07 18:58:32	1.1.2.18
@@ -522,8 +522,8 @@
 		}
 	} else if (ru->ru_rw == RU_RECOVER) {
 		DMINFO("Attempt to mark a region " SECTOR_FORMAT 
-		      ", which is being recovered.",
-		      lr->u.lr_region);
+		      "/%s which is being recovered.",
+		       lr->u.lr_region, lc->uuid + (strlen(lc->uuid) - 8));
 		DMINFO("Current recoverer: %u", ru->ru_nodeid);
 		DMINFO("Mark requester   : %u", who);
 
@@ -534,8 +534,8 @@
 	} else {
 		DMWARN("Attempt to mark a already marked region (%u,"
 		       SECTOR_FORMAT
-		       ")",
-		       who, lr->u.lr_region);
+		       "/%s)",
+		       who, lr->u.lr_region, lc->uuid + (strlen(lc->uuid) - 8));
 		mempool_free(new, region_user_pool);
 	}
 
@@ -595,8 +595,6 @@
 
 
 static int server_complete_resync_work(struct log_c *lc, struct log_request *lr, int success){
-	uint32_t info;
-
 	if (lr->u.lr_region > lc->region_count) {
 		return -EINVAL;
 	}
@@ -610,30 +608,33 @@
 			lc->sync_count++;
 		}
 	} else if (log_test_bit(lc->sync_bits, lr->u.lr_region)) {
-		DMERR("complete_resync_work region going out-of-sync: disk failure");
 		/* gone for now: lc->sync_count--; */
 		log_clear_bit(lc, lc->sync_bits, lr->u.lr_region);
 	}
 
-	info = (uint32_t)(lc->region_count - lc->sync_count);
-
-	if((info < 10001 && !(info%1000)) ||
-	   (info < 1000 && !(info%100)) ||
-	   (info < 200 && !(info%25)) ||
-	   (info < 6)){
-		DMDEBUG(SECTOR_FORMAT " out-of-sync regions remaining for %s.",
-		       lc->region_count - lc->sync_count,
-		       lc->uuid + (strlen(lc->uuid) - 8));
-	}
 	return 0;
 }
 
 
 static int server_get_sync_count(struct log_c *lc, struct log_request *lr){
+	if (lc->sync_count > lc->region_count) {
+		DMERR("sync_count (" SECTOR_FORMAT ") > region_count (" SECTOR_FORMAT ") in %s!",
+		      lc->sync_count, lc->region_count, lc->uuid + (strlen(lc->uuid) - 8));
+		disk_resume(lc);
+	}
+
 	lr->u.lr_region_rtn = lc->sync_count;
 	return 0;
 }
 
+static void report_duplicate_log(struct log_c *lc)
+{
+	DMERR("HEY!!! There are two matches for %s",
+	      lc->uuid + (strlen(lc->uuid) - 8));
+	list_for_each_entry(lc, &log_list_head, log_list) {
+		DMERR("  %s", lc->uuid + (strlen(lc->uuid) - 8));
+	}
+}
 
 static struct log_c *get_log_context(char *uuid){
 	struct log_c *lc, *r = NULL;
@@ -641,8 +642,7 @@
 	list_for_each_entry(lc, &log_list_head, log_list){
 		if(!strncmp(lc->uuid, uuid, MAX_NAME_LEN)){
 			if (r)
-				DMERR("HEY!!! There are two matches for %s",
-				      lc->uuid + (strlen(lc->uuid) - 8));
+				report_duplicate_log(lc);
 			else
 				r = lc;
 		}
@@ -932,6 +932,7 @@
 				break;
 			}
 			error = server_mark_region(lc, &lr, nodeid);
+			lr.u.lr_int_rtn = 0;
 			break;
 		case LRT_CLEAR_REGION:
 			if(!(nodeid =