db_upgrader crash when upgrade storage data

[HarrisChu]

1. use copy v2.6.1 cluster
2. replace bin folder
3. start metad
4. run command

nebula1/bin/db_upgrader --src_db_path=/data/nebula/nebula_multiple_harris/nebula1/data/storage \
                --dst_db_path=/data/nebula/nebula_multiple_tmp/nebula1/data/storage/ \
                --upgrade_meta_server=192.168.15.13:1552,192.168.15.13:2552,192.168.15.13:3552 \
                --upgrade_version=2:3

db_upgrader crashed

[vesoft@vesoft-192-168-15-13 nebula_multiple_harris]$ nebula1/bin/nebula-graphd -version
nebula-graphd version Git: a06b059a, Build Time: Jan 21 2022 22:42:26
This source code is licensed under Apache 2.0 License.

(gdb) bt
#0  0x0000000001d24fc0 in rocksdb::DBImpl::NewIterator(rocksdb::ReadOptions const&, rocksdb::ColumnFamilyHandle*) ()
#1  0x00000000013f9d7d in nebula::kvstore::RocksEngine::prefixWithoutExtractor(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, void const*, std::unique_ptr<nebula::kvstore::KVIterator, std::default_delete<nebula::kvstore::KVIterator> >*) ()
#2  0x0000000001d0a7d7 in nebula::storage::UpgraderSpace::runPartV3() ()
#3  0x00000000024429f8 in folly::ThreadPoolExecutor::runTask(std::shared_ptr<folly::ThreadPoolExecutor::Thread> const&, folly::ThreadPoolExecutor::Task&&) ()
#4  0x000000000243520f in ?? ()
#5  0x00000000024ada81 in bool folly::AtomicNotificationQueue<folly::Function<void ()> >::drive<folly::EventBase::FuncRunner&>(folly::EventBase::FuncRunner&) ()
#6  0x00000000024aeaad in non-virtual thunk to folly::EventBaseAtomicNotificationQueue<folly::Function<void ()>, folly::EventBase::FuncRunner>::handlerReady(unsigned short) ()
#7  0x0000000002569993 in ?? ()
#8  0x000000000256a067 in event_base_loop ()
#9  0x00000000024a802e in folly::EventBase::loopBody(int, bool) ()
#10 0x00000000024a891e in folly::EventBase::loop() ()
#11 0x00000000024ab558 in folly::EventBase::loopForever() ()
#12 0x0000000002435bf9 in folly::IOThreadPoolExecutor::threadRun(std::shared_ptr<folly::ThreadPoolExecutor::Thread>) ()
#13 0x0000000002444167 in void folly::detail::function::FunctionTraits<void ()>::callBig<std::_Bind<void (folly::ThreadPoolExecutor::*(folly::ThreadPoolExecutor*, std::shared_ptr<folly::ThreadPoolExecutor::Thread>))(std::shared_ptr<folly::ThreadPoolExecutor::Thread>)> >(folly::detail::function::Data&) ()
#14 0x0000000002a59ea0 in execute_native_thread_routine ()
#15 0x00007fe66f616ea5 in start_thread () from /lib64/libpthread.so.0
#16 0x00007fe66f33fb0d in clone () from /lib64/libc.so.6

[HarrisChu]

so confused

[SuperYoko]

so confused

Sorry for my careless commit, prefix has 2 parameters first
I first change it to

  nebula::cpp2::ErrorCode prefix(const std::string& prefix,
                                 const void* snapshot ,
                                 std::unique_ptr<KVIterator>* iter) override;

and then change it to

  nebula::cpp2::ErrorCode prefix(const std::string& prefix,
                                 std::unique_ptr<KVIterator>* iter,
                                 const void* snapshot = nullptr) override;

and I forget to clear code in db updater. and remain it as

auto retCode = readEngine_->prefix(prefix, nullptr, &iter);

while it can still work...

I fixed it in #3785

Maybe void* is a bad idea.