From fb14012672cefd58f8aa7587152f918824ad8ea8 Mon Sep 17 00:00:00 2001
From: Michael Vrable <mvrable@cs.ucsd.edu>
Date: Mon, 9 Aug 2010 16:00:57 -0700
Subject: [PATCH] Split cloud log segments into modestly-sized chunks.

---
 bluesky/cloudlog.c | 93 ++++++++++++++++++++++++----------------------
 bluesky/log.c      |  4 +-
 2 files changed, 51 insertions(+), 46 deletions(-)

diff --git a/bluesky/cloudlog.c b/bluesky/cloudlog.c
index 49099e1..0fdbaac 100644
--- a/bluesky/cloudlog.c
+++ b/bluesky/cloudlog.c
@@ -13,13 +13,9 @@
 
 #include "bluesky-private.h"
 
-/* The locations hash table in the file system is used to map objects to their locations.  Objects are named using 128- */
-
-typedef struct {
-    BlueSkyCloudID id;
-
-    BlueSkyCloudPointer *cloud_loc;
-} BlueSkyLocationEntry;
+// Rough size limit for a log segment.  This is not a firm limit and there are
+// no absolute guarantees on the size of a log segment.
+#define CLOUDLOG_SEGMENT_SIZE (4 << 20)
 
 BlueSkyCloudID bluesky_cloudlog_new_id()
 {
@@ -178,6 +174,22 @@ struct log_footer {
     struct logref refs[0];
 };
 
+/* Ensure that a cloud log item is loaded in memory, and if not read it in.
+ * TODO: Make asynchronous, and make this also fetch from the cloud.  Right now
+ * we only read from the log.  Log item must be locked. */
+void bluesky_cloudlog_fetch(BlueSkyCloudLog *log)
+{
+    if (log->data != NULL)
+        return;
+
+    g_assert(log->location_flags & CLOUDLOG_JOURNAL);
+
+    log->data = bluesky_log_map_object(log->fs->log, log->log_seq,
+                                       log->log_offset, log->log_size);
+
+    g_cond_broadcast(log->cond);
+}
+
 BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
                                                BlueSkyCloudLogState *state)
 {
@@ -229,6 +241,32 @@ static void find_inodes(gpointer key, gpointer value, gpointer user_data)
     state->inode_list = g_list_prepend(state->inode_list, item);
 }
 
+/* Finish up a partially-written cloud log segment and flush it to storage. */
+void bluesky_cloudlog_flush(BlueSkyFS *fs)
+{
+    BlueSkyCloudLogState *state = fs->log_state;
+    if (state->data == NULL || state->data->len == 0)
+        return;
+
+    /* TODO: Append some type of commit record to the log segment? */
+
+    g_print("Serializing %zd bytes of data to cloud\n", state->data->len);
+
+    BlueSkyStoreAsync *async = bluesky_store_async_new(fs->store);
+    async->op = STORE_OP_PUT;
+    async->key = g_strdup_printf("log-%08d-%08d",
+                                 state->location.directory,
+                                 state->location.sequence);
+    async->data = bluesky_string_new_from_gstring(state->data);
+    bluesky_store_async_submit(async);
+    bluesky_store_async_wait(async);
+    bluesky_store_async_unref(async);
+
+    state->location.sequence++;
+    state->location.offset = 0;
+    state->data = g_string_new("");
+}
+
 void bluesky_cloudlog_write_log(BlueSkyFS *fs)
 {
     BlueSkyCloudLogState *state = fs->log_state;
@@ -245,45 +283,10 @@ void bluesky_cloudlog_write_log(BlueSkyFS *fs)
         bluesky_cloudlog_unref(log);
         state->inode_list = g_list_delete_link(state->inode_list,
                                                state->inode_list);
-    }
 
-    if (state->data->len > 0) {
-        g_print("Serialized %zd bytes of data to cloud\n", state->data->len);
-
-        BlueSkyStoreAsync *async = bluesky_store_async_new(fs->store);
-        async->op = STORE_OP_PUT;
-        async->key = g_strdup_printf("log-%08d-%08d",
-                                     state->location.directory,
-                                     state->location.sequence);
-        async->data = bluesky_string_new_from_gstring(state->data);
-        bluesky_store_async_submit(async);
-        bluesky_store_async_wait(async);
-        bluesky_store_async_unref(async);
-
-        state->location.sequence++;
-        state->location.offset = 0;
-    } else {
-        g_string_free(state->data, TRUE);
+        if (state->data->len > CLOUDLOG_SEGMENT_SIZE)
+            bluesky_cloudlog_flush(fs);
     }
 
-    state->data = NULL;
-}
-
-/* Ensure that a cloud log item is loaded in memory, and if not read it in.
- * TODO: Make asynchronous, and make this also fetch from the cloud.  Right now
- * we only read from the log.  Log item must be locked. */
-void bluesky_cloudlog_fetch(BlueSkyCloudLog *log)
-{
-    if (log->data != NULL)
-        return;
-
-    g_print("Re-mapping log entry %d/%d/%d...\n",
-            log->log_seq, log->log_offset, log->log_size);
-
-    g_assert(log->location_flags & CLOUDLOG_JOURNAL);
-
-    log->data = bluesky_log_map_object(log->fs->log, log->log_seq,
-                                       log->log_offset, log->log_size);
-
-    g_cond_broadcast(log->cond);
+    bluesky_cloudlog_flush(fs);
 }
diff --git a/bluesky/log.c b/bluesky/log.c
index cec46cb..9dfaa5b 100644
--- a/bluesky/log.c
+++ b/bluesky/log.c
@@ -36,7 +36,7 @@
 
 // Rough size limit for a log segment.  This is not a firm limit and there are
 // no absolute guarantees on the size of a log segment.
-#define LOG_SEGMENT_SIZE (1 << 23)
+#define LOG_SEGMENT_SIZE (1 << 24)
 
 #define HEADER_MAGIC 0x676f4c0a
 #define FOOTER_MAGIC 0x2e435243
@@ -295,6 +295,8 @@ BlueSkyRCStr *bluesky_log_map_object(BlueSkyLog *log,
 
         g_hash_table_insert(log->mmap_cache, GINT_TO_POINTER(log_seq), map);
 
+        g_print("Mapped log segment %d...\n", log_seq);
+
         close(fd);
     }
 
-- 
2.20.1