[sheepdog] [PATCH v2 2/4] sheepfs: implement the http interface of sheepfs by using libcurl
Robin Dong
robin.k.dong at gmail.com
Fri Jan 24 04:47:01 CET 2014
From: Robin Dong <sanbai at taobao.com>
Use libcurl to call the interface of http for object-storage and implement
the read/write operation.
Signed-off-by: Robin Dong <sanbai at taobao.com>
---
sheepfs/Makefile.am | 2 +-
sheepfs/core.c | 1 +
sheepfs/http.c | 268 +++++++++++++++++++++++++++++++++++++++++++++++++++-
sheepfs/sheepfs.h | 5 +
4 files changed, 274 insertions(+), 2 deletions(-)
diff --git a/sheepfs/Makefile.am b/sheepfs/Makefile.am
index 64f84f2..a5e84ba 100644
--- a/sheepfs/Makefile.am
+++ b/sheepfs/Makefile.am
@@ -26,7 +26,7 @@ sbin_PROGRAMS = sheepfs
sheepfs_SOURCES = core.c cluster.c vdi.c shadow_file.c volume.c node.c \
config.c http.c
-sheepfs_LDADD = ../lib/libsheepdog.a $(fuse_LIBS) $(LIBS) -lpthread
+sheepfs_LDADD = ../lib/libsheepdog.a $(fuse_LIBS) $(LIBS) -lpthread -lcurl
sheepfs_DEPENDENCIES = ../lib/libsheepdog.a
noinst_HEADERS = sheepfs.h
diff --git a/sheepfs/core.c b/sheepfs/core.c
index 2f900bc..f80270b 100644
--- a/sheepfs/core.c
+++ b/sheepfs/core.c
@@ -79,6 +79,7 @@ static struct sheepfs_file_operation {
[OP_HTTP_ADDRESS] = { http_address_read, http_address_write,
http_address_get_size },
[OP_HTTP_OBJECT] = { NULL, http_object_write },
+ [OP_OBJECT] = { object_read, NULL, object_get_size },
};
__printf(3, 4)
diff --git a/sheepfs/http.c b/sheepfs/http.c
index e220e9c..e3521bd 100644
--- a/sheepfs/http.c
+++ b/sheepfs/http.c
@@ -18,6 +18,7 @@
#include <stdlib.h>
#include <stdio.h>
#include <time.h>
+#include <curl/curl.h>
#include "strbuf.h"
#include "sheepfs.h"
@@ -27,6 +28,9 @@
#define PATH_HTTP_ADDRESS "/http/address"
#define PATH_HTTP_OBJECT "/http/object"
+#define HTTP_SIZE_NAME "user.object.size"
+#define HTTP_SIZE_SIZE sizeof(uint64_t)
+
int create_http_layout(void)
{
if (shadow_dir_create(PATH_HTTP) < 0)
@@ -64,8 +68,270 @@ size_t http_address_get_size(const char *path)
return 0;
}
+static uint64_t curl_get_object_size(const char *url)
+{
+ CURL *curl;
+ CURLcode res;
+ double content_length;
+
+ curl = curl_easy_init();
+ if (!curl) {
+ sheepfs_pr("Failed to init curl");
+ goto out;
+ }
+
+ curl_easy_setopt(curl, CURLOPT_FAILONERROR, true);
+ curl_easy_setopt(curl, CURLOPT_NOBODY, true);
+ curl_easy_setopt(curl, CURLOPT_CUSTOMREQUEST, "HEAD");
+ curl_easy_setopt(curl, CURLOPT_URL, url);
+ res = curl_easy_perform(curl);
+ if (res == CURLE_OK) {
+ res = curl_easy_getinfo(curl, CURLINFO_CONTENT_LENGTH_DOWNLOAD,
+ &content_length);
+ if (res != CURLE_OK) {
+ sheepfs_pr("Failed to get size of object %s",
+ curl_easy_strerror(res));
+ content_length = 0;
+ }
+ }
+out:
+ curl_easy_cleanup(curl);
+ return (uint64_t)content_length;
+}
+
+struct buffer_s {
+ char *mem;
+ size_t current_size;
+ size_t total_size;
+};
+
+static size_t write_cb(void *contents, size_t size, size_t nmemb, void *userp)
+{
+ size_t real_size = size * nmemb;
+ struct buffer_s *buff = (struct buffer_s *)userp;
+
+ if ((buff->current_size + real_size) > buff->total_size)
+ real_size = buff->total_size - buff->current_size;
+ memcpy(buff->mem + buff->current_size, contents, real_size);
+ buff->current_size += real_size;
+ return real_size;
+}
+
+static size_t curl_read_object(const char *url, char *buf, size_t size,
+ off_t offset)
+{
+ CURL *curl;
+ CURLcode res;
+ char header[PATH_MAX];
+ double content_length;
+ struct buffer_s buffer = { 0 };
+ struct curl_slist *headers = NULL;
+
+ curl = curl_easy_init();
+ if (!curl) {
+ sheepfs_pr("Failed to init curl");
+ goto out;
+ }
+
+ snprintf(header, sizeof(header), "Range: bytes=%"PRIu64"-%"PRIu64,
+ offset, offset + size - 1);
+ headers = curl_slist_append(headers, header);
+ curl_easy_setopt(curl, CURLOPT_FAILONERROR, true);
+ curl_easy_setopt(curl, CURLOPT_URL, url);
+ curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
+ curl_easy_setopt(curl, CURLOPT_CUSTOMREQUEST, "GET");
+ curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_cb);
+ buffer.mem = buf;
+ buffer.total_size = size;
+ curl_easy_setopt(curl, CURLOPT_WRITEDATA, &buffer);
+ res = curl_easy_perform(curl);
+ if (res == CURLE_OK) {
+ res = curl_easy_getinfo(curl, CURLINFO_CONTENT_LENGTH_DOWNLOAD,
+ &content_length);
+ if (res != CURLE_OK) {
+ sheepfs_pr("Failed to getinfo res: %s",
+ curl_easy_strerror(res));
+ size = 0;
+ goto out;
+ }
+ if ((size_t)content_length > size) {
+ sheepfs_pr("Failed to get correct CONTENT_LENGTH, "
+ "content_length: %"PRIu64", get_size: %"PRIu64,
+ (size_t)content_length, size);
+ size = 0;
+ }
+ } else {
+ sheepfs_pr("Failed to call libcurl res: %s, url: %s",
+ curl_easy_strerror(res), url);
+ size = 0;
+ }
+out:
+ curl_slist_free_all(headers);
+ curl_easy_cleanup(curl);
+ return size;
+}
+
+static bool curl_object_exists(const char *url)
+{
+ CURL *curl;
+ CURLcode res;
+ bool ret = false;
+
+ curl = curl_easy_init();
+ if (!curl) {
+ sheepfs_pr("Failed to init curl");
+ goto out;
+ }
+
+ curl_easy_setopt(curl, CURLOPT_FAILONERROR, true);
+ curl_easy_setopt(curl, CURLOPT_NOBODY, true);
+ curl_easy_setopt(curl, CURLOPT_CUSTOMREQUEST, "HEAD");
+ curl_easy_setopt(curl, CURLOPT_URL, url);
+ res = curl_easy_perform(curl);
+ if (res == CURLE_OK)
+ ret = true;
+ else
+ sheepfs_pr("Failed to call libcurl res: %s, url: %s",
+ curl_easy_strerror(res), url);
+
+out:
+ curl_easy_cleanup(curl);
+ return ret;
+}
+
+struct object_cache {
+ uint64_t offset;
+ uint64_t length;
+ unsigned char *buffer;
+};
+
+static int generate_url(const char *buff, int size, char *url, int url_len)
+{
+ char address[PATH_MAX], *ch;
+ int len, ret = 0;
+
+ len = shadow_file_read(PATH_HTTP_ADDRESS, address, sizeof(address), 0);
+ if (len <= 0) {
+ sheepfs_pr("Can't get address of http");
+ ret = -EINVAL;
+ goto out;
+ }
+
+ /* remove last '\n' of address */
+ ch = strchr(address, '\n');
+ if (ch != NULL)
+ *ch = '\0';
+
+ snprintf(url, url_len, "http://%s/v1%.*s", address, size, buff);
+out:
+ return ret;
+}
+
+int object_read(const char *path, char *buf, size_t size, off_t offset)
+{
+ char url[PATH_MAX];
+ char *pos;
+ int ret;
+
+ pos = strstr(path, PATH_HTTP);
+ if (!pos) {
+ sheepfs_pr("Invalid Path %s", path);
+ ret = -EINVAL;
+ goto out;
+ }
+
+ pos += strlen(PATH_HTTP);
+ /* don't need '\n' at the end of 'path' */
+ ret = generate_url(pos, strlen(path) - strlen(PATH_HTTP),
+ url, PATH_MAX);
+ if (ret)
+ goto out;
+
+ ret = curl_read_object(url, buf, size, offset);
+out:
+ return ret;
+}
+
+size_t object_get_size(const char *path)
+{
+ uint64_t object_size;
+ if (shadow_file_getxattr(path, HTTP_SIZE_NAME, &object_size,
+ HTTP_SIZE_SIZE) < 0)
+ return 0;
+
+ return object_size;
+}
+
+static int object_create_entry(const char *entry, const char *url)
+{
+ struct strbuf buf = STRBUF_INIT;
+ char *args[3], path[PATH_MAX];
+ int nr, ret = -EINVAL;
+ uint64_t size;
+
+ nr = split_path(entry, ARRAY_SIZE(args), args);
+ if (nr != ARRAY_SIZE(args)) {
+ sheepfs_pr("Invalid argument %d, %s", nr, entry);
+ goto out;
+ }
+
+ strbuf_addf(&buf, "%s", PATH_HTTP);
+ for (int i = 0; i < nr; i++) {
+ strbuf_addf(&buf, "/%s", args[i]);
+ snprintf(path, sizeof(path), "%.*s", (int)buf.len, buf.buf);
+ if (i == (nr - 1)) {
+ if (shadow_file_create(path) < 0) {
+ sheepfs_pr("Create file %s fail", path);
+ goto out;
+ }
+ size = curl_get_object_size(url);
+ if (size <= 0) {
+ sheepfs_pr("Failed to get size of object");
+ shadow_file_delete(path);
+ goto out;
+ }
+ if (shadow_file_setxattr(path, HTTP_SIZE_NAME, &size,
+ HTTP_SIZE_SIZE) < 0) {
+ sheepfs_pr("Failed to setxattr for %s",
+ HTTP_SIZE_NAME);
+ shadow_file_delete(path);
+ goto out;
+ }
+ if (sheepfs_set_op(path, OP_OBJECT) < 0) {
+ sheepfs_pr("Set_op %s fail", path);
+ shadow_file_delete(path);
+ goto out;
+ }
+ } else {
+ if (shadow_dir_create(path) < 0) {
+ sheepfs_pr("Create dir %s fail", path);
+ goto out;
+ }
+ }
+ }
+ ret = 0;
+out:
+ strbuf_release(&buf);
+ return ret;
+}
+
int http_object_write(const char *path, const char *buf, size_t size,
off_t ignore)
{
- return size;
+ char entry[PATH_MAX], url[PATH_MAX];
+ int ret = -EINVAL;
+
+ /* don't need '\n' at the end of 'buf' */
+ ret = generate_url(buf, size - 1, url, PATH_MAX);
+ if (ret)
+ goto out;
+
+ if (curl_object_exists(url)) {
+ snprintf(entry, size, "%s", buf);
+ ret = object_create_entry(entry, url);
+ if (ret >= 0)
+ ret = size;
+ }
+out:
+ return ret;
}
diff --git a/sheepfs/sheepfs.h b/sheepfs/sheepfs.h
index 64e560a..b426f66 100644
--- a/sheepfs/sheepfs.h
+++ b/sheepfs/sheepfs.h
@@ -17,6 +17,7 @@ enum sheepfs_opcode {
OP_VOLUME,
OP_HTTP_ADDRESS,
OP_HTTP_OBJECT,
+ OP_OBJECT,
};
#define COMMAND_LEN 512
@@ -107,4 +108,8 @@ int http_address_write(const char *path, const char *buf, size_t size,
size_t http_address_get_size(const char *path);
int http_object_write(const char *path, const char *buf, size_t size,
off_t ignore);
+
+int object_read(const char *path, char *buf, size_t size, off_t ignore);
+size_t object_get_size(const char *path);
+int object_open(const char *path, struct fuse_file_info *);
#endif
--
1.7.12.4
More information about the sheepdog
mailing list