Skip to content

Implemented recover_file. #209

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
jessicayu626 wants to merge 1 commit into
base: develop
Choose a base branch
from
Open

Implemented recover_file. #209

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions zookeeper/include/zk_nn_client.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -266,6 +266,7 @@ class ZkNnClient : public ZkClientCommon {
RenewLeaseResponseProto &res);
void recover_lease(RecoverLeaseRequestProto &req,
RecoverLeaseResponseProto &res);
void recover_file(std::string &file_path);
/**
* Get info of the file.
* @param req GetFileInfoRequestProto
Expand Down
77 changes: 73 additions & 4 deletions zookeeper/source/zk_nn_client.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@
#include <google/protobuf/message.h>
#include <erasurecoding.pb.h>
#include <zk_dn_client.h>
#include <zk_nn_client.h>


using hadoop::hdfs::AddBlockRequestProto;
Expand Down Expand Up @@ -380,6 +381,77 @@ void ZkNnClient::renew_lease(RenewLeaseRequestProto &req,
}
}

void ZkNnClient::recover_file(std::string &file_path) {
int error_code;
FileZNode znode_data;
read_file_znode(znode_data, file_path);
std::uint64_t file_length = znode_data.length;

std::vector<std::string> children;
if (!zk->get_children(ZookeeperBlocksPath(file_path), children, error_code)) {
LOG(FATAL) << "[recover_file] Failed to get children for " << file_path;
return;
}

std::uint64_t size_left = file_length;
std::sort(children.begin(), children.end());
for (auto &child : children) {
std::vector<std::uint8_t> block_vec;
std::uint64_t block_id;
auto child_path = util::concat_path(ZookeeperBlocksPath(file_path), child);
if (!zk->get(child_path, block_vec, error_code, sizeof(block_id))) {
LOG(ERROR) << "[recover_file] failed block retrieval";
}
block_id = *reinterpret_cast<std::uint64_t *>(block_vec.data());
std::string block_metadata_path = get_block_metadata_path(block_id);
BlockZNode block_data;
std::vector<std::uint8_t> block_data_vec(sizeof(block_data));
if (!zk->get(block_metadata_path, block_data_vec, error_code,
sizeof(block_data))) {
LOG(ERROR) << "[recover_file] Failed getting block size for "
<< block_metadata_path << " with error " << error_code;
}
memcpy(&block_data, &block_data_vec[0], sizeof(block_data));
uint64_t block_size = block_data.block_size;

if (size_left < block_size) {
Copy link
Copy Markdown
Contributor

@Antzen Antzen Nov 27, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this case would ever be executed under our current implementation of file appends, but it's probably good to keep if future implementations would use "reserve blocks" for file mutability.

// Updates the block_size
block_data.block_size = size_left;
memcpy(&block_data_vec[0], &block_data, sizeof(block_data));
if (!zk->set(block_metadata_path, block_data_vec, error_code,
sizeof(block_data))) {
LOG(ERROR) << "[recover_file] Failed setting block size for "
<< block_metadata_path << " with error " << error_code;
}
size_left = 0;
} else if (size_left <= 0) {
// Removes the block
std::vector<std::string> datanodes;

if (!zk->get_children(get_block_metadata_path(block_id),
datanodes, error_code)) {
LOG(ERROR) << "[recover_file] Failed getting datanode "
"locations for block: "
<< block_id
<< " with error: "
<< error_code;
return;
}
// push delete to datanodes' delete queues
for (auto &dn : datanodes) {
auto delete_queue = util::concat_path(DELETE_QUEUES, dn);
auto delete_item = util::concat_path(delete_queue, "block-");
LOG(INFO) << "[recover_file] pushed to delete queue " << delete_item;
if (!zk->create(delete_item, block_vec, error_code, false)) {
LOG(INFO) << "[recover_file] pushed create " << delete_item;
}
blockDeleted(block_id, dn);
Copy link
Copy Markdown
Contributor

@Antzen Antzen Nov 27, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This looks good to me, but we should double check that this is a valid delete process with some additional unit tests.

}
}
size_left -= block_size;
}
}

void ZkNnClient::recover_lease(RecoverLeaseRequestProto &req,
RecoverLeaseResponseProto &res) {
std::string client_name = req.clientname();
Expand Down Expand Up @@ -421,9 +493,7 @@ void ZkNnClient::recover_lease(RecoverLeaseRequestProto &req,
}
std::string lease_holder_client = children[0];
if (lease_expired(lease_holder_client)) {
// TODO(elfyingying): start the lease recovery process
// that closes the file for the
// client and make sure the replicas are consistent.
recover_file(file_path);
if (!zk->delete_node(LeaseZookeeperPath(file_path) + '/' +
lease_holder_client, error_code, true)) {
LOG(ERROR) << "[recover_lease] Failed to delete the dead lease holder "
Expand Down Expand Up @@ -928,7 +998,6 @@ ZkNnClient::DeleteResponse ZkNnClient::destroy_helper(const std::string &path,
}
for (auto &child : children) {
auto child_path = util::concat_path(ZookeeperBlocksPath(path), child);
child_path = child_path;
ops.push_back(zk->build_delete_op(child_path));
std::vector<std::uint8_t> block_vec;
std::uint64_t block;
Expand Down