[Tarantool-patches] [PATCH 06/13] wal: encapsulate ER_WAL_IO

Vladislav Shpilevoy v.shpilevoy at tarantool.org
Sat Jun 12 00:56:14 MSK 2021


ER_WAL_IO is set on any WAL error if it was after journal_write()
success. It is not correct, because there can be plenty of
reasons.

In WAL it could be an actual IO error or a cascading rollback in
progress. When used for transactions, it could be an error related
to synchronous transactions like a timeout, or a persistent
ROLLBACK.

These errors are overridden by ER_WAL_IO. The patch encapsulates
the diag installation for bad journal write and for transaction
rollback.

The next patches are going to introduce more error codes and use
proper ones to install a diag.

Part of #6027
---
 src/box/applier.cc  | 10 +++++-----
 src/box/journal.c   |  8 ++++++++
 src/box/journal.h   | 10 ++++++++++
 src/box/raft.c      |  2 +-
 src/box/txn.c       |  8 +++++++-
 src/box/txn.h       | 10 ++++++++++
 src/box/txn_limbo.c |  2 +-
 7 files changed, 42 insertions(+), 8 deletions(-)

diff --git a/src/box/applier.cc b/src/box/applier.cc
index 60d648795..3fd71393d 100644
--- a/src/box/applier.cc
+++ b/src/box/applier.cc
@@ -711,7 +711,7 @@ applier_read_tx(struct applier *applier, struct stailq *rows, double timeout)
 }
 
 static void
-applier_rollback_by_wal_io(void)
+applier_rollback_by_wal_io(int64_t signature)
 {
 	/*
 	 * Setup shared applier diagnostic area.
@@ -725,7 +725,7 @@ applier_rollback_by_wal_io(void)
 	 * rollback may happen a way later after it was passed to
 	 * the journal engine.
 	 */
-	diag_set(ClientError, ER_WAL_IO);
+	diag_set_txn_sign(signature);
 	diag_set_error(&replicaset.applier.diag,
 		       diag_last_error(diag_get()));
 
@@ -747,7 +747,7 @@ applier_txn_rollback_cb(struct trigger *trigger, void *event)
 	 * special handling.
 	 */
 	if (txn->signature != TXN_SIGNATURE_SYNC_ROLLBACK)
-		applier_rollback_by_wal_io();
+		applier_rollback_by_wal_io(txn->signature);
 	return 0;
 }
 
@@ -787,7 +787,7 @@ apply_synchro_row_cb(struct journal_entry *entry)
 	struct synchro_entry *synchro_entry =
 		(struct synchro_entry *)entry->complete_data;
 	if (entry->res < 0) {
-		applier_rollback_by_wal_io();
+		applier_rollback_by_wal_io(entry->res);
 	} else {
 		txn_limbo_process(&txn_limbo, synchro_entry->req);
 		trigger_run(&replicaset.applier.on_wal_write, NULL);
@@ -838,7 +838,7 @@ apply_synchro_row(struct xrow_header *row)
 	if (journal_write(&entry.base) != 0)
 		goto err;
 	if (entry.base.res < 0) {
-		diag_set(ClientError, ER_WAL_IO);
+		diag_set_journal_res(entry.base.res);
 		goto err;
 	}
 	return 0;
diff --git a/src/box/journal.c b/src/box/journal.c
index df491610a..0a1e9932a 100644
--- a/src/box/journal.c
+++ b/src/box/journal.c
@@ -31,6 +31,7 @@
 #include "journal.h"
 #include <small/region.h>
 #include <diag.h>
+#include "error.h"
 
 struct journal *current_journal = NULL;
 
@@ -41,6 +42,13 @@ struct journal_queue journal_queue = {
 	.waiter_count = 0,
 };
 
+void
+diag_set_journal_res_detailed(const char *file, unsigned line, int64_t res)
+{
+	(void)res;
+	diag_set_detailed(file, line, ClientError, ER_WAL_IO);
+}
+
 struct journal_entry *
 journal_entry_new(size_t n_rows, struct region *region,
 		  journal_write_async_f write_async_cb,
diff --git a/src/box/journal.h b/src/box/journal.h
index 4ab7e8afb..01ea60f72 100644
--- a/src/box/journal.h
+++ b/src/box/journal.h
@@ -44,6 +44,16 @@ struct journal_entry;
 
 typedef void (*journal_write_async_f)(struct journal_entry *entry);
 
+/**
+ * Convert a result of a journal entry write to an error installed into the
+ * current diag.
+ */
+void
+diag_set_journal_res_detailed(const char *file, unsigned line, int64_t res);
+
+#define diag_set_journal_res(res)						\
+	diag_set_journal_res_detailed(__FILE__, __LINE__, res)
+
 /**
  * An entry for an abstract journal.
  * Simply put, a write ahead log request.
diff --git a/src/box/raft.c b/src/box/raft.c
index 55dee4cb1..7f787c0c5 100644
--- a/src/box/raft.c
+++ b/src/box/raft.c
@@ -312,7 +312,7 @@ box_raft_write(struct raft *raft, const struct raft_msg *msg)
 	if (is_err)
 		goto fail;
 	if (entry->res < 0) {
-		diag_set(ClientError, ER_WAL_IO);
+		diag_set_journal_res(entry->res);
 		goto fail;
 	}
 
diff --git a/src/box/txn.c b/src/box/txn.c
index 761630939..ac11127d3 100644
--- a/src/box/txn.c
+++ b/src/box/txn.c
@@ -248,6 +248,12 @@ txn_free(struct txn *txn)
 	stailq_add(&txn_cache, &txn->in_txn_cache);
 }
 
+void
+diag_set_txn_sign_detailed(const char *file, unsigned line, int64_t signature)
+{
+	return diag_set_journal_res_detailed(file, line, signature);
+}
+
 struct txn *
 txn_begin(void)
 {
@@ -906,7 +912,7 @@ txn_commit(struct txn *txn)
 	if (journal_write(req) != 0)
 		goto rollback_io;
 	if (req->res < 0) {
-		diag_set(ClientError, ER_WAL_IO);
+		diag_set_journal_res(req->res);
 		goto rollback_io;
 	}
 	if (txn_has_flag(txn, TXN_WAIT_SYNC)) {
diff --git a/src/box/txn.h b/src/box/txn.h
index a06aaea23..d51761bc9 100644
--- a/src/box/txn.h
+++ b/src/box/txn.h
@@ -123,6 +123,16 @@ enum {
 	TXN_SIGNATURE_SYNC_ROLLBACK = -3,
 };
 
+/**
+ * Convert a result of a transaction execution to an error installed into the
+ * current diag.
+ */
+void
+diag_set_txn_sign_detailed(const char *file, unsigned line, int64_t signature);
+
+#define diag_set_txn_sign(signature)						\
+	diag_set_txn_sign_detailed(__FILE__, __LINE__, signature)
+
 /**
  * Status of a transaction.
  */
diff --git a/src/box/txn_limbo.c b/src/box/txn_limbo.c
index 83b86387c..b03c71514 100644
--- a/src/box/txn_limbo.c
+++ b/src/box/txn_limbo.c
@@ -338,7 +338,7 @@ txn_limbo_write_synchro(struct txn_limbo *limbo, uint16_t type, int64_t lsn,
 	if (journal_write(entry) != 0)
 		goto fail;
 	if (entry->res < 0) {
-		diag_set(ClientError, ER_WAL_IO);
+		diag_set_journal_res(entry->res);
 		goto fail;
 	}
 	return;
-- 
2.24.3 (Apple Git-128)



More information about the Tarantool-patches mailing list