sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-01-10 01:18:15 +03:00
Code

ctdb-vacuum: Fix the incorrect counting of remote errors

Browse Source 
If a node fails to delete a record in TRY_DELETE_RECORDS control during
vacuuming, then it's possible that other nodes also may fail to delete a
record.  So instead of deleting the record from RB tree on first failure,
keep track of the remote failures.

Update delete_list.remote_error and delete_list.left statistics only
once per record during the delete_record_traverse.

BUG: https://bugzilla.samba.org/show_bug.cgi?id=13641

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>
This commit is contained in:
Amitay Isaacs 2018-02-14 15:18:17 +11:00 committed by Amitay Isaacs
parent 202b9027ba
commit ef05239717
1 changed files with 13 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
ctdb/server/ctdb_vacuum.c
View File

@ -107,6 +107,7 @@ struct delete_record_data {
struct ctdb_context *ctdb;
struct ctdb_db_context *ctdb_db;
struct ctdb_ltdb_header hdr;
uint32_t remote_fail_count;
TDB_DATA key;
uint8_t keydata[1];
};
@ -149,6 +150,7 @@ static int insert_delete_record_data_into_tree(struct ctdb_context *ctdb,
memcpy(dd->keydata, key.dptr, key.dsize);
dd->hdr = *hdr;
dd->remote_fail_count = 0;
hash = ctdb_hash(&key);
@ -451,6 +453,13 @@ static int delete_record_traverse(void *param, void *data)
uint32_t lmaster;
uint32_t hash = ctdb_hash(&(dd->key));
if (dd->remote_fail_count > 0) {
vdata->count.delete_list.remote_error++;
vdata->count.delete_list.left--;
talloc_free(dd);
return 0;
}
res = tdb_chainlock(ctdb_db->ltdb->tdb, dd->key);
if (res != 0) {
DEBUG(DEBUG_ERR,
@ -828,22 +837,17 @@ static void ctdb_process_delete_list(struct ctdb_db_context *ctdb_db,
ctdb_hash(&reckey));
if (dd != NULL) {
/*
* The other node could not delete the
* record and it is the first node that
* failed. So we should remove it from
* the tree and update statistics.
* The remote node could not delete the
* record. Since other remote nodes can
* also fail, we just mark the record.
*/
talloc_free(dd);
vdata->count.delete_list.remote_error++;
vdata->count.delete_list.left--;
dd->remote_fail_count++;
} else {
DEBUG(DEBUG_ERR, (__location__ " Failed to "
"find record with hash 0x%08x coming "
"back from TRY_DELETE_RECORDS "
"control in delete list.\n",
ctdb_hash(&reckey)));
vdata->count.delete_list.local_error++;
vdata->count.delete_list.left--;
}
rec = (struct ctdb_rec_data_old *)(rec->length + (uint8_t *)rec);