dbcheck: fixes for eliminate_orphaned_path_records and repair_bad_paths

eliminate_orphaned_path_records:
  * will no longer remove paths that a subpaths of other paths.
  * will run even if bvfs is used, as the fix described above will prevent harm to the bvfs structures.

repair_bad_paths:
  * prevents modifying the empty path entry (""). This entry is used as parent directory, even parent of "/".

src/cats/bdb_query_enum_class.h

@@ -74,6 +74,8 @@ class B_DB_QUERY_ENUM_CLASS {
       SQL_QUERY_list_volumes_by_poolid_count_1 = 68,
       SQL_QUERY_list_joblog_2 = 69,
       SQL_QUERY_list_joblog_count_1 = 70,
-      SQL_QUERY_NUMBER = 71
+      SQL_QUERY_get_orphaned_paths_0 = 71,
+      SQL_QUERY_get_bad_paths_0 = 72,
+      SQL_QUERY_NUMBER = 73
    } SQL_QUERY_ENUM;
 };

src/cats/bdb_query_names.inc

@@ -72,5 +72,7 @@ const char *B_DB::query_names[] = {
 "list_volumes_by_poolid_count_1",
 "list_joblog_2",
 "list_joblog_count_1",
+"get_orphaned_paths_0",
+"get_bad_paths_0",
 NULL
 };

src/cats/dml/0072_get_orphaned_paths_0

@@ -0,0 +1,13 @@
+#
+# dbcheck: eliminate_orphaned_path_records
+#
+# get all Paths not used by Files
+# and are not substrings to others Paths.
+#
+# LIKE: _% wildcard, but at least one character
+#
+SELECT DISTINCT ParentPath.PathId,File.PathId,ParentPath.Path
+FROM Path AS ParentPath LEFT OUTER JOIN File USING(PathId)
+WHERE File.PathId IS NULL
+  AND NOT EXISTS (SELECT 1 FROM Path WHERE Path LIKE ParentPath.Path || '_%%')
+LIMIT 300000

src/cats/dml/0072_get_orphaned_paths_0.mysql

@@ -0,0 +1,13 @@
+#
+# dbcheck: eliminate_orphaned_path_records
+#
+# get all Paths not used by Files
+# and are not substrings to others Paths.
+#
+# LIKE: _% wildcard, but at least one character
+#
+SELECT DISTINCT ParentPath.PathId,File.PathId,ParentPath.Path
+FROM Path AS ParentPath LEFT OUTER JOIN File USING(PathId)
+WHERE File.PathId IS NULL
+  AND NOT EXISTS (SELECT 1 FROM Path WHERE Path LIKE CONCAT(ParentPath.Path,'_%%'))
+LIMIT 300000

src/cats/dml/0073_get_bad_paths_0

@@ -0,0 +1,7 @@
+#
+# dbcheck: repair_bad_paths
+#
+SELECT PathId, Path FROM Path
+WHERE  Path != ''
+  AND  Path NOT LIKE '%%/'
+

src/cats/dml/README

@@ -9,6 +9,8 @@ The .inc files only need to be updated when the queries
 are altered or new ones are added.
 
 NOTE:
+  * Naming: Number_Description_NumberOfArguments
+  * As % are used for replace variables, % in SQL LIKE statements must be escaped (% => %%).
   * each query must consist of a single SQL statement,
     as more statement, separated by columns (";") need special result handling for MySQL.
 

src/cats/mysql_queries.inc

@@ -912,5 +912,19 @@ const char *B_DB_MYSQL::query_definitions[] = {
 "WHERE Log.JobId=%s "
 ,
 
+/* 0072_get_orphaned_paths_0.mysql */
+"SELECT DISTINCT ParentPath.PathId,File.PathId,ParentPath.Path "
+"FROM Path AS ParentPath LEFT OUTER JOIN File USING(PathId) "
+"WHERE File.PathId IS NULL "
+  "AND NOT EXISTS (SELECT 1 FROM Path WHERE Path LIKE CONCAT(ParentPath.Path,'_%%')) "
+"LIMIT 300000 "
+,
+
+/* 0073_get_bad_paths_0 */
+"SELECT PathId, Path FROM Path "
+"WHERE  Path != '' "
+  "AND  Path NOT LIKE '%%/' "
+,
+
 NULL
 };

src/cats/postgresql_queries.inc

@@ -888,5 +888,19 @@ const char *B_DB_POSTGRESQL::query_definitions[] = {
 "WHERE Log.JobId=%s "
 ,
 
+/* 0072_get_orphaned_paths_0 */
+"SELECT DISTINCT ParentPath.PathId,File.PathId,ParentPath.Path "
+"FROM Path AS ParentPath LEFT OUTER JOIN File USING(PathId) "
+"WHERE File.PathId IS NULL "
+  "AND NOT EXISTS (SELECT 1 FROM Path WHERE Path LIKE ParentPath.Path || '_%%') "
+"LIMIT 300000 "
+,
+
+/* 0073_get_bad_paths_0 */
+"SELECT PathId, Path FROM Path "
+"WHERE  Path != '' "
+  "AND  Path NOT LIKE '%%/' "
+,
+
 NULL
 };

src/cats/sqlite_queries.inc

@@ -897,5 +897,19 @@ const char *B_DB_SQLITE::query_definitions[] = {
 "WHERE Log.JobId=%s "
 ,
 
+/* 0072_get_orphaned_paths_0 */
+"SELECT DISTINCT ParentPath.PathId,File.PathId,ParentPath.Path "
+"FROM Path AS ParentPath LEFT OUTER JOIN File USING(PathId) "
+"WHERE File.PathId IS NULL "
+  "AND NOT EXISTS (SELECT 1 FROM Path WHERE Path LIKE ParentPath.Path || '_%%') "
+"LIMIT 300000 "
+,
+
+/* 0073_get_bad_paths_0 */
+"SELECT PathId, Path FROM Path "
+"WHERE  Path != '' "
+  "AND  Path NOT LIKE '%%/' "
+,
+
 NULL
 };

src/dird/dbcheck.c

@@ -710,15 +710,9 @@ static void eliminate_orphaned_file_records()
 static void eliminate_orphaned_path_records()
 {
    db_int64_ctx lctx;
+   POOL_MEM query(PM_MESSAGE);
 
    lctx.count = 0;
-   db->sql_query("SELECT 1 FROM Job WHERE HasCache=1 LIMIT 1", db_int64_handler, &lctx);
-   if (lctx.count > 0) {
-      printf(_("Pruning orphaned Path entries isn't possible when using BVFS.\n"));
-      fflush(stdout);
-      return;
-   }
-
    idx_tmp_name = NULL;
    /*
     * Check the existence of the required "one column" index
@@ -732,16 +726,14 @@ static void eliminate_orphaned_path_records()
       }
    }
 
-   const char *query = "SELECT DISTINCT Path.PathId,File.PathId FROM Path "
-               "LEFT OUTER JOIN File USING(PathId) "
-               "WHERE File.PathId IS NULL LIMIT 300000";
+   db->fill_query(query, B_DB::SQL_QUERY_get_orphaned_paths_0);
 
    printf(_("Checking for orphaned Path entries. This may take some time!\n"));
    if (verbose > 1) {
-      printf("%s\n", query);
+      printf("%s\n", query.c_str());
    }
    fflush(stdout);
-   if (!make_id_list(query, &id_list)) {
+   if (!make_id_list(query.c_str(), &id_list)) {
       exit(1);
    }
    /*
@@ -769,7 +761,7 @@ static void eliminate_orphaned_path_records()
       } else {
          break;                       /* get out if not updating db */
       }
-      if (!make_id_list(query, &id_list)) {
+      if (!make_id_list(query.c_str(), &id_list)) {
          exit(1);
       }
    }
@@ -1059,18 +1051,16 @@ static void repair_bad_filenames()
 
 static void repair_bad_paths()
 {
-   const char *query;
+   POOL_MEM query(PM_MESSAGE);
    int i;
 
    printf(_("Checking for Paths without a trailing slash\n"));
-   query = "SELECT PathId,Path from Path "
-           "WHERE  Path != '' "
-           "AND    Path NOT LIKE '%/'";
+   db->fill_query(query, B_DB::SQL_QUERY_get_bad_paths_0);
    if (verbose > 1) {
-      printf("%s\n", query);
+      printf("%s\n", query.c_str());
    }
    fflush(stdout);
-   if (!make_id_list(query, &id_list)) {
+   if (!make_id_list(query.c_str(), &id_list)) {
       exit(1);
    }
    printf(_("Found %d bad Path records.\n"), id_list.num_ids);