consistent hash utility

Signed-off-by: Aakash Arayambeth <[email protected]>

bbinc/consistent_hash.h

@@ -0,0 +1,101 @@
+#ifndef CH_HASH_H
+#define CH_HASH_H
+#include <stddef.h>
+#include <stdint.h>
+#include <openssl/bn.h>
+#include <list.h>
+typedef struct consistent_hash ch_hash_t;
+typedef struct consistent_hash_node ch_hash_node_t;
+typedef int (*hash_func)(uint8_t *, size_t, uint32_t *);
+enum ch_err {
+    CH_NOERR = 0,
+    CH_ERR_PARAM = 1,
+    CH_ERR_MALLOC = 2,
+    CH_ERR_HASH = 3,
+    CH_ERR_DUP = 4
+};
+
+struct consistent_hash_node {
+    uint8_t *data;
+    size_t data_len;
+    LINKC_T(struct consistent_hash_node) lnk;
+};
+
+typedef struct consistent_hash_keyhash {
+    uint64_t hash_val; 
+    struct consistent_hash_node *node;
+}ch_keyhash_t;
+
+struct consistent_hash {
+    LISTC_T(struct consistent_hash_node) nodes;         /* List of nodes for key hashes to refer to */
+    uint32_t num_nodes;
+    struct consistent_hash_keyhash **key_hashes;   /* Array of key hashes of nodes and (optionally)
+                                                      their copies. I went with an array over a hash
+                                                      since we need this list to be sorted in order to look 
+                                                      up the next largest keyhash that has an associated node,
+                                                      given a keyhash*/
+    uint32_t num_keyhashes;
+    pthread_rwlock_t lock;
+    hash_func func;
+};
+
+/*
+ * Create an in-mem abstraction of a consistent hash.
+ * num_copies -> number of copies of node to be added to the hash
+ *               Db servers, cache servers are examples of a 'Node' 
+ * ch_func -> pointer to the hash function to be used 
+ *            It takes a byte array and it's length as the input
+ *            The hash is available in the OUT parameter hash as a BIGNUM
+ *            www.openssl.org/docs/man1.0.2/man3/bn.html
+ */
+ch_hash_t *ch_hash_create(uint64_t num_copies, hash_func func);
+
+/*
+ * Add a node to the consistent hash.
+ * name -> a byte array that represents the name of the node
+ *         in the calling program.
+ * name_len -> length of above name
+ */
+int ch_hash_add_node(ch_hash_t *hash, uint8_t *name, size_t name_len, uint64_t hashval);
+
+
+/*
+ * Add a replica for a node at hashval 
+ * node_name    -> node name for which replica is being added 
+ * node_len     -> size (in bytes) of node_name
+*/
+int ch_hash_add_replica(ch_hash_t *hash, uint8_t *node_name, size_t node_len, uint64_t hashval);
+
+
+/*
+ * Remove a node from the consistent hash.
+ * name -> a byte array that represents the name of the node
+ *         in the calling program.
+ * name_len -> length of above name
+ */
+int ch_hash_remove_node(ch_hash_t *hash, uint8_t *name, size_t name_len);
+
+/*
+ * Given a key and it's length, return the node on the consistent hash 
+ * that the inputs hash onto
+ */
+ch_hash_node_t *ch_hash_find_node(ch_hash_t *hash, uint8_t *key, size_t key_len);
+
+/*
+ * Free all memory associated with a consistent hash
+ */
+void ch_hash_free(ch_hash_t *hash);
+
+/*
+ * SHA256 and MD5 based hash funcs
+ */
+int ch_hash_sha(uint8_t *buf, size_t buf_len, uint32_t *hash);
+int ch_hash_md5(uint8_t *buf, size_t buf_len, uint32_t *hash);
+
+
+/*
+ * TEST HELPERS
+ */
+uint8_t *get_node_data(ch_hash_node_t *ch);
+size_t get_node_data_len(ch_hash_node_t *ch);
+#endif

tests/consistent_hash.test/Makefile

@@ -0,0 +1,12 @@
+ifeq ($(TESTSROOTDIR),)
+  include ../testcase.mk
+else
+  include $(TESTSROOTDIR)/testcase.mk
+endif
+ifeq ($(TEST_TIMEOUT),)
+	export TEST_TIMEOUT=1m
+endif
+
+# this is a local test, don't need cluster
+unexport CLUSTER
+export COMDB2_UNITTEST=1

tests/consistent_hash.test/README

@@ -0,0 +1 @@
+This test exercises the consistent hash utility

tests/consistent_hash.test/runit

@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+
+set -e
+set -x
+
+echo run executable that tests the consistent hash utility
+${TESTSBUILDDIR}/test_consistent_hash
+

tests/consistent_hash_bench.test/Makefile

@@ -0,0 +1,12 @@
+ifeq ($(TESTSROOTDIR),)
+  include ../testcase.mk
+else
+  include $(TESTSROOTDIR)/testcase.mk
+endif
+ifeq ($(TEST_TIMEOUT),)
+	export TEST_TIMEOUT=1m
+endif
+
+# this is a local test, don't need cluster
+unexport CLUSTER
+export COMDB2_UNITTEST=1

tests/consistent_hash_bench.test/README

@@ -0,0 +1,5 @@
+This test runs record distribution benchmark against the consistent hash utility
+It creates 5 tables ( to simulate 5 shards) and inserts a million records into these 5 tables.
+The table in which a record goes is decided by the consistent hash and the value of the field in the record
+being inserted. Next, one of the tables(shards) is dropped, which causes a certain number of records to be 
+re-distributed. This test checks the distribution before dropping a table and after dropping a table

tests/consistent_hash_bench.test/runit

@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+
+set -e
+set -x
+
+echo run executable that tests the consistent hash utility
+# s -> number of shards
+# c -> number of copies(replicas) for each shard in the consistent hash
+# r -> number of records to be inserted
+${TESTSBUILDDIR}/test_consistent_hash_bench -s 5 -c 512 -r 1000000
+

tests/tools/CMakeLists.txt

@@ -70,13 +70,17 @@ add_exe(sqlite_clnt sqlite_clnt.c)
 add_exe(ssl_multi_certs_one_process ssl_multi_certs_one_process.c)
 add_exe(stepper stepper.c stepper_client.c)
 add_exe(test_threadpool test_threadpool.c)
+add_exe(test_consistent_hash test_consistent_hash.c)
+add_exe(test_consistent_hash_bench test_consistent_hash_bench.c)
 add_exe(updater updater.c testutil.c)
 add_exe(utf8 utf8.c)
 add_exe(verify_atomics_work verify_atomics_work.c)
 
 target_link_libraries(cson_test cson)
 target_link_libraries(stepper util mem dlmalloc util)
 target_link_libraries(test_threadpool util mem dlmalloc util)
+target_link_libraries(test_consistent_hash util mem dlmalloc util)
+target_link_libraries(test_consistent_hash_bench util mem dlmalloc util)
 
 list(APPEND common-deps
   ${READLINE_LIBRARIES}
@@ -87,6 +91,7 @@ list(APPEND common-deps
   ${UNWIND_LIBRARY}
   ${CMAKE_DL_LIBS}
   m
+  crc32c
 )
 
 foreach(executable ${test-tools})