thebestvibedev/tromcho.net
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Treat copied buffers as written only after completing the write in client

Browse Source 
SYNC operation fsyncs only completed operations, so treating writes as "eligible
for fsync" before actually completing them is incorrect

It affected SCHEME=ec test_heal.sh (with immediate_commit=none) test - it was
flapping with lost writes - some non-fsynced writes were legitimately lost by
the OSD, but weren't repeated by the client
This commit is contained in:
Vitaliy Filippov
2024-06-20 02:11:53 +03:00
parent e400a851f4
commit ea0d72289c
5 changed files with 34 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/client/cluster_client_impl.h
+2 -1
View File
@@ -46,11 +46,12 @@ public:
bool is_left_merged(dirty_buf_it_t dirty_it);
bool is_right_merged(dirty_buf_it_t dirty_it);
bool is_merged(const dirty_buf_it_t & dirty_it);
void copy_write(cluster_op_t *op, int state);
void copy_write(cluster_op_t *op, int state, uint64_t new_flush_id = 0);
int repeat_ops_for(cluster_client_t *cli, osd_num_t peer_osd, pool_id_t pool_id, pg_num_t pg_num);
void start_writebacks(cluster_client_t *cli, int count);
bool read_from_cache(cluster_op_t *op, uint32_t bitmap_granularity);
void flush_buffers(cluster_client_t *cli, dirty_buf_it_t from_it, dirty_buf_it_t to_it);
void mark_flush_written(uint64_t inode, uint64_t offset, uint64_t len, uint64_t flush_id);
void fsync_start();
void fsync_error();
void fsync_ok();