gvsafronov/futriix
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Handle race where a different thread cancels the replication but we haven't closed the connection

Browse Source 
Former-commit-id: 92e48f071792985c2fb7f4c722329e832cd4c77b
This commit is contained in:
John Sully 2021-05-30 02:08:02 +00:00
parent 24377fa8fc
commit 542a83edec
1 changed files with 4 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/replication.cpp
View File

@ -1993,6 +1993,10 @@ void readSyncBulkPayload(connection *conn) {
// Should we update our database, or create from scratch? // Should we update our database, or create from scratch?
int fUpdate = g_pserver->fActiveReplica || g_pserver->enable_multimaster; int fUpdate = g_pserver->fActiveReplica || g_pserver->enable_multimaster;
redisMaster *mi = (redisMaster*)connGetPrivateData(conn); redisMaster *mi = (redisMaster*)connGetPrivateData(conn);
if (mi == nullptr) {
// We're about to be free'd so bail out
return;
}
serverAssert(GlobalLocksAcquired()); serverAssert(GlobalLocksAcquired());
serverAssert(mi->master == nullptr); serverAssert(mi->master == nullptr);