[sheepdog] [PATCH v1] sheep/http: fix error in bucket_delete

Liu Yuan namei.unix at gmail.com
Wed Aug 6 13:02:48 CEST 2014


On Tue, Aug 05, 2014 at 08:33:46PM +0800, Bingpeng Zhu wrote:
> From: NankaiZBP <nkuzbp at foxmail.com>
> 
> In current implementation, When we create a bucket, we decide
> the bnode's location in account VDI using sd_hash(bucket_name)
> as key. We handle hash conflict by linear probing hash table.
> Here is the bug:
> When we delete a bucket, we can't discard its bnode. Because
> bnode_lookup() need it to find if some bucket exists or not
> by checking adjacent bnodes. Therefore, we just zero its
> bnode.name when client want to delete a bucket. When we create
> a bucket later, we can reuse the deleted bnode if they hash to
> the same location in account VDI.
> 
> Signed-off-by: Yu Fang <bingpeng.zbp at alibaba-inc.com>
> ---
>  sheep/http/kv.c |   37 +++++++++++++++++++++++++++++--------
>  1 files changed, 29 insertions(+), 8 deletions(-)
> 
> diff --git a/sheep/http/kv.c b/sheep/http/kv.c
> index 5c4b386..2060bb1 100644
> --- a/sheep/http/kv.c
> +++ b/sheep/http/kv.c
> @@ -247,18 +247,21 @@ int kv_delete_account(struct http_request *req, const char *account)
>   */
>  
>  static int bnode_do_create(struct kv_bnode *bnode, struct sd_inode *inode,
> -			   uint32_t idx)
> +			   uint32_t idx, bool create)
>  {
>  	uint32_t vid = inode->vdi_id;
>  	uint64_t oid = vid_to_data_oid(vid, idx);
>  	int ret;
>  
>  	bnode->oid = oid;
> -	ret = sd_write_object(oid, (char *)bnode, sizeof(*bnode), 0, true);
> +	ret = sd_write_object(oid, (char *)bnode, sizeof(*bnode), 0, create);
>  	if (ret != SD_RES_SUCCESS) {
>  		sd_err("failed to create object, %" PRIx64, oid);
>  		goto out;
>  	}
> +	if (!create)
> +		goto out;
> +
>  	sd_inode_set_vid(inode, idx, vid);
>  	ret = sd_inode_write_vid(inode, idx, vid, vid, 0, false, false);
>  	if (ret != SD_RES_SUCCESS) {
> @@ -276,6 +279,7 @@ static int bnode_create(struct kv_bnode *bnode, uint32_t account_vid)
>  	uint32_t tmp_vid, idx;
>  	uint64_t hval, i;
>  	int ret;
> +	bool create = true;
>  
>  	ret = sd_read_object(vid_to_vdi_oid(account_vid), (char *)inode,
>  			     sizeof(*inode), 0);
> @@ -289,16 +293,27 @@ static int bnode_create(struct kv_bnode *bnode, uint32_t account_vid)
>  	for (i = 0; i < MAX_DATA_OBJS; i++) {
>  		idx = (hval + i) % MAX_DATA_OBJS;
>  		tmp_vid = sd_inode_get_vid(inode, idx);
> -		if (tmp_vid)
> -			continue;
> -		else
> +		if (tmp_vid) {
> +			uint64_t oid = vid_to_data_oid(account_vid, idx);
> +			char name[SD_MAX_BUCKET_NAME] = { };
> +
> +			ret = sd_read_object(oid, name, sizeof(name), 0);
> +			if (ret != SD_RES_SUCCESS)
> +				goto out;
> +			if (name[0] == 0) {
> +				create = false;
> +				goto create;
> +			}
> +		} else
>  			break;
>  	}
>  	if (i == MAX_DATA_OBJS) {
>  		ret = SD_RES_NO_SPACE;
>  		goto out;
>  	}
> -	ret = bnode_do_create(bnode, inode, idx);
> +
> +create:
> +	ret = bnode_do_create(bnode, inode, idx, create);
>  out:
>  	free(inode);
>  	return ret;
> @@ -423,6 +438,7 @@ static int bucket_delete(const char *account, uint32_t avid, const char *bucket)
>  	struct kv_bnode bnode;
>  	char onode_name[SD_MAX_VDI_LEN];
>  	char alloc_name[SD_MAX_VDI_LEN];
> +	char name[SD_MAX_BUCKET_NAME] = {};
>  	int ret;
>  
>  	snprintf(onode_name, SD_MAX_VDI_LEN, "%s/%s", account, bucket);
> @@ -436,9 +452,14 @@ static int bucket_delete(const char *account, uint32_t avid, const char *bucket)
>  	if (bnode.object_count > 0)
>  		return SD_RES_VDI_NOT_EMPTY;
>  
> -	ret = sd_discard_object(bnode.oid);
> +	/*
> +	 * We can't discard bnode because bnode_lookup() need it to find
> +	 * if some bucket exists or not by checking adjacent bnodes.
> +	 * So we just zero bnode.name to indicate a deleted bucket.
> +	 */
> +	ret = sd_write_object(bnode.oid, name, sizeof(name), 0, false);
>  	if (ret != SD_RES_SUCCESS) {
> -		sd_err("failed to discard bnode for %s", bucket);
> +		sd_err("failed to zero bnode for %s", bucket);
>  		return ret;
>  	}
>  	sd_delete_vdi(onode_name);

If we try to mimic the onode lookup algorithm as your patch did, we should also
modify the bnode_lookup() too. See onode_lookup_nolock().

Thanks
Yuan



More information about the sheepdog mailing list