Skip to content

Try to fix diskchunk flush #3166

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

Try to fix diskchunk flush #3166

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
22 changes: 22 additions & 0 deletions src/searchdreplication.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -923,6 +923,17 @@ bool HandleCmdReplicated ( RtAccum_t & tAcc )
if ( tAcc.m_dCmd.IsEmpty() )
return TlsMsg::Err ( "empty accumulator" );

// Increment active replication operations counter if the index is an RT index
auto pIndex = (RtIndex_c*)tAcc.m_pIndex;
if ( pIndex )
pIndex->m_iActiveReplicationOperations.fetch_add(1, std::memory_order_release);

// Ensure we decrement the counter when we leave this function
auto tCleanup = AtScopeExit([pIndex]() {
if ( pIndex )
pIndex->m_iActiveReplicationOperations.fetch_sub(1, std::memory_order_release);
});

const ReplicationCommand_t & tCmd = *tAcc.m_dCmd[0];
bool bCmdCluster = ( tAcc.m_dCmd.GetLength()==1
&& ( tCmd.m_eCommand==ReplCmd_e::CLUSTER_ALTER_ADD || tCmd.m_eCommand==ReplCmd_e::CLUSTER_ALTER_DROP ) );
Expand Down Expand Up @@ -1130,6 +1141,17 @@ static bool HandleCmdReplicateImpl ( RtAccum_t & tAcc, int * pDeletedCount, CSph
{
TRACE_CORO ( "sph", "HandleCmdReplicateImpl" );

// Increment active replication operations counter if the index is an RT index
auto pIndex = (RtIndex_c*)tAcc.m_pIndex;
if ( pIndex )
pIndex->m_iActiveReplicationOperations.fetch_add(1, std::memory_order_release);

// Ensure we decrement the counter when we leave this function
auto tCleanup = AtScopeExit([pIndex]() {
if ( pIndex )
pIndex->m_iActiveReplicationOperations.fetch_sub(1, std::memory_order_release);
});

CommitMonitor_c tMonitor ( tAcc, pDeletedCount, pWarning, pUpdated );

// with cluster path
Expand Down
10 changes: 10 additions & 0 deletions src/sphinxrt.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1453,6 +1453,7 @@ class RtIndex_c final : public RtIndex_i, public ISphNoncopyable, public ISphWor
mutable DWORD m_uDiskAttrStatus = 0;
std::atomic<int64_t> m_tmDataWriten = 0;
std::atomic<int64_t> m_tmDataSearched = 0;
std::atomic<int> m_iActiveReplicationOperations = 0; // Counter for active replication operations

bool m_bKeywordDict;
int m_iWordsCheckpoint = RTDICT_CHECKPOINT_V5;
Expand Down Expand Up @@ -4250,6 +4251,15 @@ bool RtIndex_c::SaveDiskChunk ( bool bForced, bool bEmergent ) REQUIRES ( m_tWor

assert ( Coro::CurrentScheduler() == m_tWorkers.SerialChunkAccess() );

// Skip saving if replication operations are in progress and this is not a forced save
// This prevents data loss in clustered environments where replication operations
// might not have fully propagated before disk chunks are saved
if ( m_iActiveReplicationOperations.load(std::memory_order_acquire) > 0 && !bForced )
{
RTSAVELOG << "SaveDiskChunk skipped due to active replication operations";
return true;
}

RTSAVELOG << "SaveDiskChunk (" << ( bForced ? "forced, " : "not forced, " ) << ( bEmergent ? "emergent, " : "not emergent " ) << ")";

m_tNSavesNow.Wait ( [] ( int iVal ) { return iVal < SIMULTANEOUS_SAVE_LIMIT; } );
Expand Down
Loading