search

apache / horaedb

Apache HoraeDB (incubating) is a high-performance, distributed, cloud native time-series database.
https://horaedb.apache.org
Apache License 2.0
2.63k stars 206 forks source link

OB_ERR_PRIMARY_KEY_DUPLICATE when doing compaction #1205

Closed jiacai2050 closed 3 weeks ago

jiacai2050 commented 1 year ago

Describe this problem

In our deployments with OBKV as wal implementation, we found following errors:

2023-09-11 10:24:03.322 ERRO [proxy/src/grpc/write.rs:33] Failed to handle write, err:Internal error, msg:Failed to execute interpreter, err:Failed to execute insert, err:Failed to write table, err:Failed to wri
te tables, table:adrtbcoreV2__DEFAULT__1__DEFAULT_goc_ctr_latency_us_bucket, err:Failed to flush table, table:adrtbcoreV2__DEFAULT__1__DEFAULT_goc_ctr_latency_us_bucket, err:Background flush failed, cannot write
 more data, retry_count:20, err:Failed to run flush job, msg:Some("table:adrtbcoreV2__DEFAULT__1__DEFAULT_goc_ctr_latency_us_bucket, table_id:4449, request_id:607920880"), err:Failed to alloc file id, err:Failed
 to alloc file id, err:Failed to write update to wal, err:Failed to write log entries, err:Failed to write table unit, namespace:manifest, wal location:WalLocation { region_id: 99, table_id: 4449 }, err:Failed t
o write log to table, region_id:99, err:Failed to write to table, table:wal_manifest_permanent_000099, err:Common error, code:ObException(OB_ERR_PRIMARY_KEY_DUPLICATE), err:OBKV server return exception in batch
response: ObTableOperationResult { base: BasePayLoad { channel_id: 1262076206, version: 1, timeout: 10000, flag: 7 }, header: ObTableResult { base: BasePayLoad { channel_id: 1262076207, version: 1, timeout: 1000
0, flag: 7 }, errorno: -5024, sql_state: [0], msg: [0] }, operation_type: Insert, entity: ObTableEntity { base: BasePayLoad { channel_id: 1262076208, version: 1, timeout: 10000, flag: 7 }, row_key: ObRowKey { ke
ys: [] }, properties: {} }, affected_rows: 0 }..                                                                               
Backtrace:     
0 <snafu::backtrace_shim::Backtrace as snafu::GenerateBacktrace>::generate::h02217cd51b8458df                                                              
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/snafu-0.6.10/src/backtrace_shim.rs:15     
  <table_kv::obkv::WriteTable<__T0> as snafu::IntoError<table_kv::obkv::Error>>::into_error::hb2712bb104fe5bf5
  /home/db/ceresdb/components/table_kv/src/obkv.rs:35                                                  
  <core::result::Result<T,E> as snafu::ResultExt<T,E>>::context::{{closure}}::ha945fa61a9d0fdc4         
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/snafu-0.6.10/src/lib.rs:318               
  core::result::Result<T,E>::map_err::h2c67e337ccf5ca8b                                                                                 
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/result.rs:860                        
  <core::result::Result<T,E> as snafu::ResultExt<T,E>>::context::h79ba9da8f5cc9087                                 
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/snafu-0.6.10/src/lib.rs:318                            
1 <table_kv::obkv::ObkvImpl as table_kv::TableKv>::write::h4c7f29ac2567028a                                          
  /home/db/ceresdb/components/table_kv/src/obkv.rs:503                             
2 wal::table_kv_impl::table_unit::TableUnitWriter::write_log::{{closure}}::{{closure}}::h374597f853fb1ebd          
  /home/db/ceresdb/wal/src/table_kv_impl/table_unit.rs:972                                           
  <tokio::runtime::blocking::task::BlockingTask<T> as core::future::future::Future>::poll::hd732123b7ce277e0                                                                                                         /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/blocking/task.rs:42                
  tokio::runtime::task::core::Core<T,S>::poll::{{closure}}::h7bc8dab5bff1edd8                                                                                                                                        /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/core.rs:311                                                                                                                 tokio::loom::std::unsafe_cell::UnsafeCell<T>::with_mut::h874e5b1272fa4233                                                                                                                                       
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/loom/std/unsafe_cell.rs:14                                                                                                               tokio::runtime::task::core::Core<T,S>::poll::h004175d454d51de2                                                                                                                                                     /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/core.rs:300                                                                                                                 tokio::runtime::task::harness::poll_future::{{closure}}::h214bee1116474500                                             
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/harness.rs:476      
  <core::panic::unwind_safe::AssertUnwindSafe<F> as core::ops::function::FnOnce<()>>::call_once::h8283513884f3bd25       
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/panic/unwind_safe.rs:271            
  std::panicking::try::do_call::h35e5c17abc1297f6                                                                         
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/panicking.rs:483                  
  std::panicking::try::h3476acde4fde0c11                                                                                           
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/panicking.rs:447 
  std::panic::catch_unwind::hc0a7aa6a07741dbc
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/panic.rs:140
  tokio::runtime::task::harness::poll_future::h4d9ff312aa07f229
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/harness.rs:464
  tokio::runtime::task::harness::Harness<T,S>::poll_inner::h4a3e33931b374525
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/harness.rs:198
  tokio::runtime::task::harness::Harness<T,S>::poll::h33185797ff335d5d
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/harness.rs:152
  tokio::runtime::task::raw::poll::hee93371fc98f2eb6
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/raw.rs:276
3 tokio::runtime::task::raw::RawTask::poll::heabf9c70e6fd77a4
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/raw.rs:200
  tokio::runtime::task::UnownedTask<S>::run::h71abe9846161356c
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/task/mod.rs:437
  tokio::runtime::blocking::pool::Task::run::h447d1827b63320cd
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/blocking/pool.rs:159
  tokio::runtime::blocking::pool::Inner::run::h512f4d4d040e4562
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/blocking/pool.rs:513
  tokio::runtime::blocking::pool::Spawner::spawn_thread::{{closure}}::h136aab7e70ce0684
  /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/tokio-1.29.1/src/runtime/blocking/pool.rs:471
  std::sys_common::backtrace::__rust_begin_short_backtrace::h0fdc9691673882ce
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/sys_common/backtrace.rs:121
4 std::thread::Builder::spawn_unchecked_::{{closure}}::{{closure}}::hafa9124fedea5fb8
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/thread/mod.rs:558
  <core::panic::unwind_safe::AssertUnwindSafe<F> as core::ops::function::FnOnce<()>>::call_once::hd98e2f10cf2e1f38
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/panic/unwind_safe.rs:271
  std::panicking::try::do_call::h46b8f1ab94cefb1f
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/panicking.rs:483
  std::panicking::try::h347f95a1f7845419
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/panicking.rs:447
  std::panic::catch_unwind::h5dd7ee5b5feeb457
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/panic.rs:140
  std::thread::Builder::spawn_unchecked_::{{closure}}::h2561cf64a40e49e6
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/thread/mod.rs:557
  core::ops::function::FnOnce::call_once{{vtable.shim}}::h46fa823051c99d1e
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/ops/function.rs:250
5 <alloc::boxed::Box<F,A> as core::ops::function::FnOnce<Args>>::call_once::hc8beb91c5e39b692
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/alloc/src/boxed.rs:1988
  <alloc::boxed::Box<F,A> as core::ops::function::FnOnce<Args>>::call_once::h20e58dce1054acc4
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/alloc/src/boxed.rs:1988
  std::sys::unix::thread::Thread::new::thread_start::h848946a57aa81736
  /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/std/src/sys/unix/thread.rs:108
6 start_thread
7 __clone
Backtrace:                                   
 0 <snafu::backtrace_shim::Backtrace as snafu::GenerateBacktrace>::generate::h02217cd51b8458df
   /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/snafu-0.6.10/src/backtrace_shim.rs:15
   <wal::manager::error::Write as snafu::IntoError<wal::manager::error::Error>>::into_error::hf5c49394fa68e61d
   /home/db/ceresdb/wal/src/manager.rs:41                                   
 1 <core::result::Result<T,E> as snafu::ResultExt<T,E>>::context::{{closure}}::h557595a434999597       
   /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/snafu-0.6.10/src/lib.rs:318
   core::result::Result<T,E>::map_err::h10c8702abcd0713f                                               
   /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/result.rs:860
   <core::result::Result<T,E> as snafu::ResultExt<T,E>>::context::h2728c3a6b3d1a399                
   /home/db/.cargo/registry/src/github.com-1ecc6299db9ec823/snafu-0.6.10/src/lib.rs:318
   <wal::table_kv_impl::wal::WalNamespaceImpl<T> as wal::manager::WalManager>::write::{{closure}}::hace34fa1b277d26d
   /home/db/ceresdb/wal/src/table_kv_impl/wal.rs:175          
 2 <core::pin::Pin<P> as core::future::future::Future>::poll::haffa190cb0c71e7d                    
   /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/future/future.rs:125
   <analytic_engine::manifest::details::WalBasedLogStore as analytic_engine::manifest::details::MetaUpdateLogStore>::append::{{closure}}::h4696a77b8bdfbd84
   /home/db/ceresdb/analytic_engine/src/manifest/details.rs:723
 3 <core::pin::Pin<P> as core::future::future::Future>::poll::h3b85d708da0a520b                         
   /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/future/future.rs:125
   analytic_engine::manifest::details::ManifestImpl::store_update_to_wal::{{closure}}::h75d609256d00d067
   /home/db/ceresdb/analytic_engine/src/manifest/details.rs:442              
   <analytic_engine::manifest::details::ManifestImpl as analytic_engine::manifest::Manifest>::apply_edit::{{closure}}::h0f6eb57b366fa8dc
   /home/db/ceresdb/analytic_engine/src/manifest/details.rs:502                      
 4 <core::pin::Pin<P> as core::future::future::Future>::poll::h6b95f43fd3e5a084    
   /rustc/11d96b59307b1702fffe871bfc2d0145d070881e/library/core/src/future/future.rs:125                          
   analytic_engine::table::data::TableData::persist_max_file_id::{{closure}}::h3408c2f626b3cd19
   /home/db/ceresdb/analytic_engine/src/table/data.rs:551
   analytic_engine::table::data::TableData::alloc_file_id::{{closure}}::{{closure}}::{{closure}}::h02b6a214ddc98868
   /home/db/ceresdb/analytic_engine/src/table/data.rs:520
   id_allocator::Inner::alloc_id::{{closure}}::h1964701c5b592986                  
   /home/db/ceresdb/components/id_allocator/src/lib.rs:52
   id_allocator::IdAllocator::alloc_id::{{closure}}::h22881461833e70d2        
   /home/db/ceresdb/components/id_allocator/src/lib.rs:80               
   analytic_engine::table::data::TableData::alloc_file_id::{{closure}}::h1c839670dd38ad8d
   /home/db/ceresdb/analytic_engine/src/table/data.rs:525                 
 5 analytic_engine::instance::flush_compaction::FlushTask::dump_normal_memtable::{{closure}}::h31660e44e945a85b
   /home/db/ceresdb/analytic_engine/src/instance/flush_compaction.rs:620                     
   analytic_engine::instance::flush_compaction::FlushTask::dump_memtables::{{closure}}::h897337f24183c2ab
   /home/db/ceresdb/analytic_engine/src/instance/flush_compaction.rs:384                     
   analytic_engine::instance::flush_compaction::FlushTask::run::{{closure}}::h6d1391310def4ea6
   /home/db/ceresdb/analytic_engine/src/instance/flush_compaction.rs:280
   analytic_engine::instance::flush_compaction::Flusher::schedule_table_flush::{{closure}}::{{closure}}::h0ce1ae1039065eee
   /home/db/ceresdb/analytic_engine/src/instance/flush_compaction.rs:252
   analytic_engine::instance::serial_executor::TableFlushScheduler::flush_sequentially::{{closure}}::{{closure}}::h1aae6e2506874b14
   /home/db/ceresdb/analytic_engine/src/instance/serial_executor.rs:219

Server version

$ ceresdb-server --version CeresDB Server Version: 1.2.6-alpha Git commit: 7f8faff79 Git branch: main Opt level: 3 Rustc version: 1.69.0-nightly Target: x86_64-unknown-linux-gnu Build date: 2023-08-17T09:04:30.077953531Z

Steps to reproduce

After some discussion with @ShiKaiWi @Rachelint, this problem will arise when close shard, and table's compaction belonging to this shard still exists, this means there will be two nodes writing the same manifest file.

Expected behavior

No error

Additional Information

In theory, when a close shard request is received, it should release all resources(such as: WAL/manifest/object_store) before finish closing shard, only when those resources all are released, then we can open this shard in a new node.

ShiKaiWi commented 1 year ago

This error still exist on the production environment. And I'll fix it in the Nov Iteration.