Skip to content
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.

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

curvefs/client: fix the data iteration error when rpc retry. #1474

Merged
merged 1 commit into from
May 23, 2022
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
4 changes: 2 additions & 2 deletions curvefs/src/client/inode_cache_manager.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -119,7 +119,7 @@ CURVEFS_ERROR InodeCacheManagerImpl::BatchGetInodeAttr(
return CURVEFS_ERROR::OK;
}

MetaStatusCode ret = metaClient_->BatchGetInodeAttr(fsId_, inodeIds, attr);
MetaStatusCode ret = metaClient_->BatchGetInodeAttr(fsId_, *inodeIds, attr);
if (MetaStatusCode::OK != ret) {
LOG(ERROR) << "metaClient BatchGetInodeAttr failed, MetaStatusCode = "
<< ret << ", MetaStatusCode_Name = "
Expand Down Expand Up @@ -150,7 +150,7 @@ CURVEFS_ERROR InodeCacheManagerImpl::BatchGetXAttr(
return CURVEFS_ERROR::OK;
}

MetaStatusCode ret = metaClient_->BatchGetXAttr(fsId_, inodeIds, xattr);
MetaStatusCode ret = metaClient_->BatchGetXAttr(fsId_, *inodeIds, xattr);
if (MetaStatusCode::OK != ret) {
LOG(ERROR) << "metaClient BatchGetXAttr failed, MetaStatusCode = "
<< ret << ", MetaStatusCode_Name = "
Expand Down
295 changes: 157 additions & 138 deletions curvefs/src/client/rpcclient/metaserver_client.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -477,14 +477,14 @@ MetaStatusCode MetaServerClientImpl::GetInode(uint32_t fsId, uint64_t inodeid,
bool GroupInodeIdByPartition(
uint32_t fsId,
std::shared_ptr<MetaCache> metaCache,
std::set<uint64_t> *inodeIds,
std::unordered_map<uint32_t, std::list<uint64_t>> *inodeGroups) {
for (const auto &it : *inodeIds) {
const std::set<uint64_t> &inodeIds,
std::unordered_map<uint32_t, std::vector<uint64_t>> *inodeGroups) {
for (const auto &it : inodeIds) {
uint32_t pId = 0;
if (metaCache->GetPartitionIdByInodeId(fsId, it, &pId)) {
auto iter = inodeGroups->find(pId);
if (iter == inodeGroups->end()) {
inodeGroups->emplace(pId, std::list<uint64_t>({it}));
inodeGroups->emplace(pId, std::vector<uint64_t>({it}));
} else {
iter->second.push_back(it);
}
Expand All @@ -497,163 +497,182 @@ bool GroupInodeIdByPartition(
return true;
}

bool MetaServerClientImpl::SplitRequestInodes(
uint32_t fsId,
const std::set<uint64_t> &inodeIds,
std::vector<std::vector<uint64_t>> *inodeGroups) {
std::unordered_map<uint32_t, std::vector<uint64_t>> groups;
bool ret = GroupInodeIdByPartition(fsId, metaCache_, inodeIds, &groups);
if (!ret) {
return false;
}
for (const auto &it : groups) {
auto iter = it.second.begin();
while (iter != it.second.end()) {
std::vector<uint64_t> tmp;
uint32_t batchLimit = opt_.batchLimit;
while (iter != it.second.end() && batchLimit > 0) {
tmp.emplace_back(*iter);
iter++;
batchLimit--;
}
inodeGroups->emplace_back(std::move(tmp));
}
}
return true;
}


MetaStatusCode MetaServerClientImpl::BatchGetInodeAttr(uint32_t fsId,
std::set<uint64_t> *inodeIds,
const std::set<uint64_t> &inodeIds,
std::list<InodeAttr> *attr) {
uint32_t limit = opt_.batchLimit;
// group inodeid by partition
std::unordered_map<uint32_t, std::list<uint64_t>> inodeGroups;
if (!GroupInodeIdByPartition(fsId, metaCache_, inodeIds, &inodeGroups)) {
// group inodeid by partition and batchlimit
std::vector<std::vector<uint64_t>> inodeGroups;
if (!SplitRequestInodes(fsId, inodeIds, &inodeGroups)) {
return MetaStatusCode::NOT_FOUND;
}

// send rpc
// TDOD(wanghai): send rpc parallelly
for (const auto &it : inodeGroups) {
auto iter = it.second.begin();
while (iter != it.second.end()) {
uint64_t inodeId = *iter;
auto task = RPCTask {
metaserverClientMetric_->batchGetInodeAttr.qps.count << 1;
LatencyUpdater updater(
&metaserverClientMetric_->batchGetInodeAttr.latency);
BatchGetInodeAttrRequest request;
BatchGetInodeAttrResponse response;
request.set_poolid(poolID);
request.set_copysetid(copysetID);
request.set_partitionid(partitionID);
request.set_fsid(fsId);
request.set_appliedindex(
metaCache_->GetApplyIndex(CopysetGroupID(poolID,
copysetID)));
uint32_t batchLimit = limit;
while (iter != it.second.end() && batchLimit > 0) {
request.add_inodeid(*iter);
iter++;
batchLimit--;
}
curvefs::metaserver::MetaServerService_Stub stub(channel);
stub.BatchGetInodeAttr(cntl, &request, &response, nullptr);

if (cntl->Failed()) {
metaserverClientMetric_->batchGetInodeAttr.eps.count << 1;
LOG(WARNING) << "BatchGetInodeAttr Failed, errorcode = "
<< cntl->ErrorCode()
<< ", error content:" << cntl->ErrorText()
<< ", log id = " << cntl->log_id();
return -cntl->ErrorCode();
}
if (it.empty()) {
LOG(WARNING) << "BatchGetInodeAttr request empty.";
return MetaStatusCode::PARAM_ERROR;
}
uint64_t inodeId = *it.begin();
auto task = RPCTask {
metaserverClientMetric_->batchGetInodeAttr.qps.count << 1;
LatencyUpdater updater(
&metaserverClientMetric_->batchGetInodeAttr.latency);
BatchGetInodeAttrRequest request;
BatchGetInodeAttrResponse response;
request.set_poolid(poolID);
request.set_copysetid(copysetID);
request.set_partitionid(partitionID);
request.set_fsid(fsId);
request.set_appliedindex(
metaCache_->GetApplyIndex(CopysetGroupID(poolID,
copysetID)));
*request.mutable_inodeid() = { it.begin(), it.end() };

curvefs::metaserver::MetaServerService_Stub stub(channel);
stub.BatchGetInodeAttr(cntl, &request, &response, nullptr);

if (cntl->Failed()) {
metaserverClientMetric_->batchGetInodeAttr.eps.count << 1;
LOG(WARNING) << "BatchGetInodeAttr Failed, errorcode = "
<< cntl->ErrorCode()
<< ", error content:" << cntl->ErrorText()
<< ", log id = " << cntl->log_id();
return -cntl->ErrorCode();
}

MetaStatusCode ret = response.statuscode();
if (ret != MetaStatusCode::OK) {
LOG_IF(WARNING, ret != MetaStatusCode::NOT_FOUND)
<< "BatchGetInodeAttr failed, errcode = " << ret
<< ", errmsg = " << MetaStatusCode_Name(ret);
} else if (response.attr_size() > 0 &&
response.has_appliedindex()) {
auto retAttr = response.attr();
for_each(retAttr.begin(), retAttr.end(),
[&](InodeAttr &a) { attr->push_back(a); });
metaCache_->UpdateApplyIndex(
CopysetGroupID(poolID, copysetID),
response.appliedindex());
} else {
LOG(WARNING) << "BatchGetInodeAttr ok, but"
<< " applyIndex or attr not set in response: "
<< response.DebugString();
return -1;
}
return ret;
};
auto taskCtx = std::make_shared<TaskContext>(
MetaServerOpType::BatchGetInodeAttr, task, fsId, inodeId);
BatchGetInodeAttrExcutor excutor(
opt_, metaCache_, channelManager_, taskCtx);
auto ret = ConvertToMetaStatusCode(excutor.DoRPCTask());
MetaStatusCode ret = response.statuscode();
if (ret != MetaStatusCode::OK) {
attr->clear();
return ret;
LOG(ERROR) << "BatchGetInodeAttr failed, errcode = " << ret
<< ", errmsg = " << MetaStatusCode_Name(ret);
} else if (response.attr_size() > 0 &&
response.has_appliedindex()) {
auto *attrs = response.mutable_attr();
attr->insert(attr->end(),
std::make_move_iterator(attrs->begin()),
std::make_move_iterator(attrs->end()));
metaCache_->UpdateApplyIndex(
CopysetGroupID(poolID, copysetID),
response.appliedindex());
} else {
LOG(WARNING) << "BatchGetInodeAttr ok, but"
<< " applyIndex or attr not set in response: "
<< response.DebugString();
return -1;
}
return ret;
};
auto taskCtx = std::make_shared<TaskContext>(
MetaServerOpType::BatchGetInodeAttr, task, fsId, inodeId);
BatchGetInodeAttrExcutor excutor(
opt_, metaCache_, channelManager_, taskCtx);
auto ret = ConvertToMetaStatusCode(excutor.DoRPCTask());
if (ret != MetaStatusCode::OK) {
attr->clear();
return ret;
}
}
return MetaStatusCode::OK;
}

MetaStatusCode MetaServerClientImpl::BatchGetXAttr(uint32_t fsId,
std::set<uint64_t> *inodeIds,
const std::set<uint64_t> &inodeIds,
std::list<XAttr> *xattr) {
uint32_t limit = opt_.batchLimit;
// group inodeid by partition
std::unordered_map<uint32_t, std::list<uint64_t>> inodeGroups;
if (!GroupInodeIdByPartition(fsId, metaCache_, inodeIds, &inodeGroups)) {
// group inodeid by partition and batchlimit
std::vector<std::vector<uint64_t>> inodeGroups;
if (!SplitRequestInodes(fsId, inodeIds, &inodeGroups)) {
return MetaStatusCode::NOT_FOUND;
}

// send rpc
// TDOD(wanghai): send rpc parallelly
for (const auto &it : inodeGroups) {
auto iter = it.second.begin();
while (iter != it.second.end()) {
uint64_t inodeId = *iter;
auto task = RPCTask {
metaserverClientMetric_->batchGetXattr.qps.count << 1;
LatencyUpdater updater(
&metaserverClientMetric_->batchGetXattr.latency);
BatchGetXAttrRequest request;
BatchGetXAttrResponse response;
request.set_poolid(poolID);
request.set_copysetid(copysetID);
request.set_partitionid(partitionID);
request.set_fsid(fsId);
request.set_appliedindex(
metaCache_->GetApplyIndex(
CopysetGroupID(poolID, copysetID)));
uint32_t batchLimit = limit;
while (iter != it.second.end() && batchLimit > 0) {
request.add_inodeid(*iter);
iter++;
batchLimit--;
}
curvefs::metaserver::MetaServerService_Stub stub(channel);
stub.BatchGetXAttr(cntl, &request, &response, nullptr);

if (cntl->Failed()) {
metaserverClientMetric_->batchGetXattr.eps.count << 1;
LOG(WARNING) << "BatchGetXAttr Failed, errorcode = "
<< cntl->ErrorCode()
<< ", error content:" << cntl->ErrorText()
<< ", log id = " << cntl->log_id();
return -cntl->ErrorCode();
}
if (it.empty()) {
LOG(WARNING) << "BatchGetInodeXAttr request empty.";
return MetaStatusCode::PARAM_ERROR;
}

MetaStatusCode ret = response.statuscode();
if (ret != MetaStatusCode::OK) {
LOG_IF(WARNING, ret != MetaStatusCode::NOT_FOUND)
<< "BatchGetXAttr failed, errcode = " << ret
<< ", errmsg = " << MetaStatusCode_Name(ret);
} else if (response.xattr_size() > 0 &&
response.has_appliedindex()) {
auto retXattr = response.xattr();
for_each(retXattr.begin(), retXattr.end(),
[&](XAttr &a) { xattr->push_back(a); });
metaCache_->UpdateApplyIndex(
CopysetGroupID(poolID, copysetID),
response.appliedindex());
} else {
LOG(WARNING) << "BatchGetXAttr ok, but"
<< " applyIndex or attr not set in response: "
<< response.DebugString();
return -1;
}
return ret;
};
auto taskCtx = std::make_shared<TaskContext>(
MetaServerOpType::BatchGetInodeAttr, task, fsId, inodeId);
BatchGetInodeAttrExcutor excutor(
opt_, metaCache_, channelManager_, taskCtx);
auto ret = ConvertToMetaStatusCode(excutor.DoRPCTask());
uint64_t inodeId = *it.begin();
auto task = RPCTask {
metaserverClientMetric_->batchGetXattr.qps.count << 1;
LatencyUpdater updater(
&metaserverClientMetric_->batchGetXattr.latency);
Comment on lines +619 to +623
Copy link
Contributor

Choose a reason for hiding this comment

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

the overall process is still serial, can we send requests parallelly?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Yes, add a TODO and improve this in later pr.

BatchGetXAttrRequest request;
BatchGetXAttrResponse response;
request.set_poolid(poolID);
request.set_copysetid(copysetID);
request.set_partitionid(partitionID);
request.set_fsid(fsId);
request.set_appliedindex(
metaCache_->GetApplyIndex(
CopysetGroupID(poolID, copysetID)));
*request.mutable_inodeid() = { it.begin(), it.end() };

curvefs::metaserver::MetaServerService_Stub stub(channel);
stub.BatchGetXAttr(cntl, &request, &response, nullptr);

if (cntl->Failed()) {
metaserverClientMetric_->batchGetXattr.eps.count << 1;
LOG(WARNING) << "BatchGetXAttr Failed, errorcode = "
<< cntl->ErrorCode()
<< ", error content:" << cntl->ErrorText()
<< ", log id = " << cntl->log_id();
return -cntl->ErrorCode();
}

MetaStatusCode ret = response.statuscode();
if (ret != MetaStatusCode::OK) {
xattr->clear();
return ret;
LOG(ERROR) << "BatchGetXAttr failed, errcode = " << ret
<< ", errmsg = " << MetaStatusCode_Name(ret);
} else if (response.xattr_size() > 0 &&
response.has_appliedindex()) {
auto *xattrs = response.mutable_xattr();
xattr->insert(xattr->end(),
std::make_move_iterator(xattrs->begin()),
std::make_move_iterator(xattrs->end()));
metaCache_->UpdateApplyIndex(
CopysetGroupID(poolID, copysetID),
response.appliedindex());
} else {
LOG(WARNING) << "BatchGetXAttr ok, but"
<< " applyIndex or attr not set in response: "
<< response.DebugString();
return -1;
}
return ret;
};
auto taskCtx = std::make_shared<TaskContext>(
MetaServerOpType::BatchGetInodeAttr, task, fsId, inodeId);
BatchGetInodeAttrExcutor excutor(
opt_, metaCache_, channelManager_, taskCtx);
auto ret = ConvertToMetaStatusCode(excutor.DoRPCTask());
if (ret != MetaStatusCode::OK) {
xattr->clear();
return ret;
}
}
return MetaStatusCode::OK;
Expand Down
Loading