From 16bc6ba2268e3660d026076264de8666356e00bf Mon Sep 17 00:00:00 2001 From: Michael Adam Date: Thu, 3 Dec 2009 17:29:54 +0100 Subject: s3:dbwrap_ctdb: start rewrite of transactions using the global lock (g_lock) This simplifies the transaction code a lot: * transaction_start essentially consists of acquiring a global lock. * No write operations at all are performed on the local database until the transaction is committed: Every store operation is just going into the marshall buffer. * The commit operation calls a new simplified TRANS3_COMMIT control in ctdb which rolls out thae changes to all nodes including the node that is performing the transaction. Michael --- source3/lib/dbwrap_ctdb.c | 480 ++++++++++++++++------------------------------ 1 file changed, 160 insertions(+), 320 deletions(-) (limited to 'source3') diff --git a/source3/lib/dbwrap_ctdb.c b/source3/lib/dbwrap_ctdb.c index 8e188d0ab5..c32398acdb 100644 --- a/source3/lib/dbwrap_ctdb.c +++ b/source3/lib/dbwrap_ctdb.c @@ -22,10 +22,10 @@ #include "ctdb.h" #include "ctdb_private.h" #include "ctdbd_conn.h" +#include "g_lock.h" struct db_ctdb_transaction_handle { struct db_ctdb_ctx *ctx; - bool in_replay; /* * we store the reads and writes done under a transaction: * - one list stores both reads and writes (m_all), @@ -35,6 +35,7 @@ struct db_ctdb_transaction_handle { struct ctdb_marshall_buffer *m_write; uint32_t nesting; bool nested_cancel; + char *lock_name; }; struct db_ctdb_ctx { @@ -42,6 +43,7 @@ struct db_ctdb_ctx { struct tdb_wrap *wtdb; uint32 db_id; struct db_ctdb_transaction_handle *transaction; + struct g_lock_ctx *lock_ctx; }; struct db_ctdb_rec { @@ -297,145 +299,21 @@ static struct ctdb_rec_data *db_ctdb_marshall_loop_next(struct ctdb_marshall_buf return r; } - -static int32_t db_ctdb_transaction_active(uint32_t db_id) -{ - int32_t status; - NTSTATUS ret; - TDB_DATA indata; - - indata.dptr = (uint8_t *)&db_id; - indata.dsize = sizeof(db_id); - - ret = ctdbd_control_local(messaging_ctdbd_connection(), - CTDB_CONTROL_TRANS2_ACTIVE, 0, 0, - indata, NULL, NULL, &status); - - if (!NT_STATUS_IS_OK(ret)) { - DEBUG(2, ("ctdb control TRANS2_ACTIVE failed\n")); - return -1; - } - - return status; -} - - /** * CTDB transaction destructor */ static int db_ctdb_transaction_destructor(struct db_ctdb_transaction_handle *h) { - tdb_transaction_cancel(h->ctx->wtdb->tdb); - return 0; -} - -/** - * start a transaction on a ctdb database: - * - lock the transaction lock key - * - start the tdb transaction - */ -static int db_ctdb_transaction_fetch_start(struct db_ctdb_transaction_handle *h) -{ - struct db_record *rh; - struct db_ctdb_rec *crec; - TDB_DATA key; - TALLOC_CTX *tmp_ctx; - const char *keyname = CTDB_TRANSACTION_LOCK_KEY; - int ret; - struct db_ctdb_ctx *ctx = h->ctx; - TDB_DATA data; - pid_t pid; NTSTATUS status; - struct ctdb_ltdb_header header; - int32_t transaction_status; - key.dptr = (uint8_t *)discard_const(keyname); - key.dsize = strlen(keyname); - -again: - tmp_ctx = talloc_new(h); - - rh = fetch_locked_internal(ctx, tmp_ctx, key, true); - if (rh == NULL) { - DEBUG(0,(__location__ " Failed to fetch_lock database\n")); - talloc_free(tmp_ctx); - return -1; - } - crec = talloc_get_type_abort(rh->private_data, struct db_ctdb_rec); - - transaction_status = db_ctdb_transaction_active(ctx->db_id); - if (transaction_status == 1) { - unsigned long int usec = (1000 + random()) % 100000; - DEBUG(3, ("Transaction already active on db_id[0x%08x]." - "Re-trying after %lu microseconds...", - ctx->db_id, usec)); - talloc_free(tmp_ctx); - usleep(usec); - goto again; - } - - /* - * store the pid in the database: - * it is not enought that the node is dmaster... - */ - pid = getpid(); - data.dptr = (unsigned char *)&pid; - data.dsize = sizeof(pid_t); - crec->header.rsn++; - crec->header.dmaster = get_my_vnn(); - status = db_ctdb_ltdb_store(ctx, key, &(crec->header), data); + status = g_lock_unlock(h->ctx->lock_ctx, h->lock_name); if (!NT_STATUS_IS_OK(status)) { - DEBUG(0, (__location__ " Failed to store pid in transaction " - "record: %s\n", nt_errstr(status))); - talloc_free(tmp_ctx); + DEBUG(0, ("g_lock_unlock failed: %s\n", nt_errstr(status))); return -1; } - - talloc_free(rh); - - ret = tdb_transaction_start(ctx->wtdb->tdb); - if (ret != 0) { - DEBUG(0,(__location__ " Failed to start tdb transaction\n")); - talloc_free(tmp_ctx); - return -1; - } - - status = db_ctdb_ltdb_fetch(ctx, key, &header, tmp_ctx, &data); - if (!NT_STATUS_IS_OK(status)) { - DEBUG(0, (__location__ " failed to refetch transaction lock " - "record inside transaction: %s - retrying\n", - nt_errstr(status))); - tdb_transaction_cancel(ctx->wtdb->tdb); - talloc_free(tmp_ctx); - goto again; - } - - if (header.dmaster != get_my_vnn()) { - DEBUG(3, (__location__ " refetch transaction lock record : " - "we are not dmaster any more " - "(dmaster[%u] != my_vnn[%u]) - retrying\n", - header.dmaster, get_my_vnn())); - tdb_transaction_cancel(ctx->wtdb->tdb); - talloc_free(tmp_ctx); - goto again; - } - - if ((data.dsize != sizeof(pid_t)) || (*(pid_t *)(data.dptr) != pid)) { - DEBUG(3, (__location__ " refetch transaction lock record: " - "another local process has started a transaction " - "(stored pid [%u] != my pid [%u]) - retrying\n", - *(pid_t *)(data.dptr), pid)); - tdb_transaction_cancel(ctx->wtdb->tdb); - talloc_free(tmp_ctx); - goto again; - } - - talloc_free(tmp_ctx); - return 0; } - /** * CTDB dbwrap API: transaction_start function * starts a transaction on a persistent database @@ -443,7 +321,7 @@ again: static int db_ctdb_transaction_start(struct db_context *db) { struct db_ctdb_transaction_handle *h; - int ret; + NTSTATUS status; struct db_ctdb_ctx *ctx = talloc_get_type_abort(db->private_data, struct db_ctdb_ctx); @@ -466,9 +344,22 @@ static int db_ctdb_transaction_start(struct db_context *db) h->ctx = ctx; - ret = db_ctdb_transaction_fetch_start(h); - if (ret != 0) { - talloc_free(h); + h->lock_name = talloc_asprintf(h, "transaction_db_0x%08x", + (unsigned int)ctx->db_id); + if (h->lock_name == NULL) { + DEBUG(0, ("talloc_asprintf failed\n")); + TALLOC_FREE(h); + return -1; + } + + /* + * Wait a day, i.e. forever... + */ + status = g_lock_lock(ctx->lock_ctx, h->lock_name, G_LOCK_WRITE, + timeval_set(86400, 0)); + if (!NT_STATUS_IS_OK(status)) { + DEBUG(0, ("g_lock_lock failed: %s\n", nt_errstr(status))); + TALLOC_FREE(h); return -1; } @@ -481,7 +372,65 @@ static int db_ctdb_transaction_start(struct db_context *db) return 0; } +static bool pull_newest_from_marshall_buffer(struct ctdb_marshall_buffer *buf, + TDB_DATA key, + struct ctdb_ltdb_header *pheader, + TALLOC_CTX *mem_ctx, + TDB_DATA *pdata) +{ + struct ctdb_rec_data *rec = NULL; + struct ctdb_ltdb_header h; + bool found; + TDB_DATA data; + int i; + + if (buf == NULL) { + return false; + } + + /* + * Walk the list of records written during this + * transaction. If we want to read one we have already + * written, return the last written sample. Thus we do not do + * a "break;" for the first hit, this record might have been + * overwritten later. + */ + + for (i=0; icount; i++) { + TDB_DATA tkey, tdata; + uint32_t reqid; + + rec = db_ctdb_marshall_loop_next(buf, rec, &reqid, &h, &tkey, + &tdata); + if (rec == NULL) { + return false; + } + + if (tdb_data_equal(key, tkey)) { + found = true; + data = tdata; + } + } + + if (!found) { + return false; + } + + if (pdata != NULL) { + data.dptr = (uint8_t *)talloc_memdup(mem_ctx, data.dptr, + data.dsize); + if ((data.dsize != 0) && (data.dptr == NULL)) { + return false; + } + *pdata = data; + } + + if (pheader != NULL) { + *pheader = h; + } + return true; +} /* fetch a record inside a transaction @@ -492,6 +441,13 @@ static int db_ctdb_transaction_fetch(struct db_ctdb_ctx *db, { struct db_ctdb_transaction_handle *h = db->transaction; NTSTATUS status; + bool found; + + found = pull_newest_from_marshall_buffer(h->m_write, key, NULL, + mem_ctx, data); + if (found) { + return 0; + } status = db_ctdb_ltdb_fetch(h->ctx, key, NULL, mem_ctx, data); @@ -501,14 +457,14 @@ static int db_ctdb_transaction_fetch(struct db_ctdb_ctx *db, return -1; } - if (!h->in_replay) { - h->m_all = db_ctdb_marshall_add(h, h->m_all, h->ctx->db_id, 1, key, NULL, *data); - if (h->m_all == NULL) { - DEBUG(0,(__location__ " Failed to add to marshalling record\n")); - data->dsize = 0; - talloc_free(data->dptr); - return -1; - } + h->m_all = db_ctdb_marshall_add(h, h->m_all, h->ctx->db_id, 1, key, + NULL, *data); + if (h->m_all == NULL) { + DEBUG(0,(__location__ " Failed to add to marshalling " + "record\n")); + data->dsize = 0; + talloc_free(data->dptr); + return -1; } return 0; @@ -543,6 +499,11 @@ static struct db_record *db_ctdb_fetch_locked_transaction(struct db_ctdb_ctx *ct result->store = db_ctdb_store_transaction; result->delete_rec = db_ctdb_delete_transaction; + if (pull_newest_from_marshall_buffer(ctx->transaction->m_write, key, + NULL, result, &result->value)) { + return result; + } + ctdb_data = tdb_fetch(ctx->wtdb->tdb, key); if (ctdb_data.dptr == NULL) { /* create the record */ @@ -619,27 +580,35 @@ static int db_ctdb_transaction_store(struct db_ctdb_transaction_handle *h, TDB_DATA key, TDB_DATA data) { TALLOC_CTX *tmp_ctx = talloc_new(h); - int ret; TDB_DATA rec; struct ctdb_ltdb_header header; - NTSTATUS status; + + ZERO_STRUCT(header); /* we need the header so we can update the RSN */ - rec = tdb_fetch(h->ctx->wtdb->tdb, key); - if (rec.dptr == NULL) { - /* the record doesn't exist - create one with us as dmaster. - This is only safe because we are in a transaction and this - is a persistent database */ - ZERO_STRUCT(header); - } else { - memcpy(&header, rec.dptr, sizeof(struct ctdb_ltdb_header)); - rec.dsize -= sizeof(struct ctdb_ltdb_header); - /* a special case, we are writing the same data that is there now */ - if (data.dsize == rec.dsize && - memcmp(data.dptr, rec.dptr + sizeof(struct ctdb_ltdb_header), data.dsize) == 0) { - SAFE_FREE(rec.dptr); - talloc_free(tmp_ctx); - return 0; + + if (!pull_newest_from_marshall_buffer(h->m_write, key, &header, + NULL, NULL)) { + + rec = tdb_fetch(h->ctx->wtdb->tdb, key); + + if (rec.dptr != NULL) { + memcpy(&header, rec.dptr, + sizeof(struct ctdb_ltdb_header)); + rec.dsize -= sizeof(struct ctdb_ltdb_header); + + /* + * a special case, we are writing the same + * data that is there now + */ + if (data.dsize == rec.dsize && + memcmp(data.dptr, + rec.dptr + sizeof(struct ctdb_ltdb_header), + data.dsize) == 0) { + SAFE_FREE(rec.dptr); + talloc_free(tmp_ctx); + return 0; + } } SAFE_FREE(rec.dptr); } @@ -647,13 +616,13 @@ static int db_ctdb_transaction_store(struct db_ctdb_transaction_handle *h, header.dmaster = get_my_vnn(); header.rsn++; - if (!h->in_replay) { - h->m_all = db_ctdb_marshall_add(h, h->m_all, h->ctx->db_id, 0, key, NULL, data); - if (h->m_all == NULL) { - DEBUG(0,(__location__ " Failed to add to marshalling record\n")); - talloc_free(tmp_ctx); - return -1; - } + h->m_all = db_ctdb_marshall_add(h, h->m_all, h->ctx->db_id, 0, key, + NULL, data); + if (h->m_all == NULL) { + DEBUG(0,(__location__ " Failed to add to marshalling " + "record\n")); + talloc_free(tmp_ctx); + return -1; } h->m_write = db_ctdb_marshall_add(h, h->m_write, h->ctx->db_id, 0, key, &header, data); @@ -663,16 +632,8 @@ static int db_ctdb_transaction_store(struct db_ctdb_transaction_handle *h, return -1; } - status = db_ctdb_ltdb_store(h->ctx, key, &header, data); - if (NT_STATUS_IS_OK(status)) { - ret = 0; - } else { - ret = -1; - } - talloc_free(tmp_ctx); - - return ret; + return 0; } @@ -708,64 +669,6 @@ static NTSTATUS db_ctdb_delete_transaction(struct db_record *rec) return NT_STATUS_OK; } - -/* - replay a transaction - */ -static int ctdb_replay_transaction(struct db_ctdb_transaction_handle *h) -{ - int ret, i; - struct ctdb_rec_data *rec = NULL; - - h->in_replay = true; - talloc_free(h->m_write); - h->m_write = NULL; - - ret = db_ctdb_transaction_fetch_start(h); - if (ret != 0) { - return ret; - } - - for (i=0;im_all->count;i++) { - TDB_DATA key, data; - - rec = db_ctdb_marshall_loop_next(h->m_all, rec, NULL, NULL, &key, &data); - if (rec == NULL) { - DEBUG(0, (__location__ " Out of records in ctdb_replay_transaction?\n")); - goto failed; - } - - if (rec->reqid == 0) { - /* its a store */ - if (db_ctdb_transaction_store(h, key, data) != 0) { - goto failed; - } - } else { - TDB_DATA data2; - TALLOC_CTX *tmp_ctx = talloc_new(h); - - if (db_ctdb_transaction_fetch(h->ctx, tmp_ctx, key, &data2) != 0) { - talloc_free(tmp_ctx); - goto failed; - } - if (data2.dsize != data.dsize || - memcmp(data2.dptr, data.dptr, data.dsize) != 0) { - /* the record has changed on us - we have to give up */ - talloc_free(tmp_ctx); - goto failed; - } - talloc_free(tmp_ctx); - } - } - - return 0; - -failed: - tdb_transaction_cancel(h->ctx->wtdb->tdb); - return -1; -} - - /* commit a transaction */ @@ -774,11 +677,8 @@ static int db_ctdb_transaction_commit(struct db_context *db) struct db_ctdb_ctx *ctx = talloc_get_type_abort(db->private_data, struct db_ctdb_ctx); NTSTATUS rets; - int ret; int status; - int retries = 0; struct db_ctdb_transaction_handle *h = ctx->transaction; - enum ctdb_controls failure_control = CTDB_CONTROL_TRANS2_ERROR; if (h == NULL) { DEBUG(0,(__location__ " transaction commit with no open transaction on db 0x%08x\n", ctx->db_id)); @@ -798,102 +698,29 @@ static int db_ctdb_transaction_commit(struct db_context *db) DEBUG(5,(__location__ " Commit transaction on db 0x%08x\n", ctx->db_id)); - talloc_set_destructor(h, NULL); - - /* our commit strategy is quite complex. - - - we first try to commit the changes to all other nodes - - - if that works, then we commit locally and we are done - - - if a commit on another node fails, then we need to cancel - the transaction, then restart the transaction (thus - opening a window of time for a pending recovery to - complete), then replay the transaction, checking all the - reads and writes (checking that reads give the same data, - and writes succeed). Then we retry the transaction to the - other nodes - */ - again: if (h->m_write == NULL) { /* no changes were made, potentially after a retry */ - tdb_transaction_cancel(h->ctx->wtdb->tdb); - talloc_free(h); - ctx->transaction = NULL; - return 0; + goto done; } /* tell ctdbd to commit to the other nodes */ - rets = ctdbd_control_local(messaging_ctdbd_connection(), - retries==0?CTDB_CONTROL_TRANS2_COMMIT:CTDB_CONTROL_TRANS2_COMMIT_RETRY, + rets = ctdbd_control_local(messaging_ctdbd_connection(), + CTDB_CONTROL_TRANS3_COMMIT, h->ctx->db_id, 0, - db_ctdb_marshall_finish(h->m_write), NULL, NULL, &status); + db_ctdb_marshall_finish(h->m_write), + NULL, NULL, &status); if (!NT_STATUS_IS_OK(rets) || status != 0) { - tdb_transaction_cancel(h->ctx->wtdb->tdb); - sleep(1); - - if (!NT_STATUS_IS_OK(rets)) { - failure_control = CTDB_CONTROL_TRANS2_ERROR; - } else { - /* work out what error code we will give if we - have to fail the operation */ - switch ((enum ctdb_trans2_commit_error)status) { - case CTDB_TRANS2_COMMIT_SUCCESS: - case CTDB_TRANS2_COMMIT_SOMEFAIL: - case CTDB_TRANS2_COMMIT_TIMEOUT: - failure_control = CTDB_CONTROL_TRANS2_ERROR; - break; - case CTDB_TRANS2_COMMIT_ALLFAIL: - failure_control = CTDB_CONTROL_TRANS2_FINISHED; - break; - } - } - - if (++retries == 100) { - DEBUG(0,(__location__ " Giving up transaction on db 0x%08x after %d retries failure_control=%u\n", - h->ctx->db_id, retries, (unsigned)failure_control)); - ctdbd_control_local(messaging_ctdbd_connection(), failure_control, - h->ctx->db_id, CTDB_CTRL_FLAG_NOREPLY, - tdb_null, NULL, NULL, NULL); - h->ctx->transaction = NULL; - talloc_free(h); - ctx->transaction = NULL; - return -1; - } - - if (ctdb_replay_transaction(h) != 0) { - DEBUG(0,(__location__ " Failed to replay transaction failure_control=%u\n", - (unsigned)failure_control)); - ctdbd_control_local(messaging_ctdbd_connection(), failure_control, - h->ctx->db_id, CTDB_CTRL_FLAG_NOREPLY, - tdb_null, NULL, NULL, NULL); - h->ctx->transaction = NULL; - talloc_free(h); - ctx->transaction = NULL; - return -1; - } + /* + * TODO: + * check the database sequence number and + * compare it to the seqnum after applying the + * marshall buffer. If it is the same: return success. + */ goto again; - } else { - failure_control = CTDB_CONTROL_TRANS2_ERROR; } - /* do the real commit locally */ - ret = tdb_transaction_commit(h->ctx->wtdb->tdb); - if (ret != 0) { - DEBUG(0,(__location__ " Failed to commit transaction failure_control=%u\n", - (unsigned)failure_control)); - ctdbd_control_local(messaging_ctdbd_connection(), failure_control, h->ctx->db_id, - CTDB_CTRL_FLAG_NOREPLY, tdb_null, NULL, NULL, NULL); - h->ctx->transaction = NULL; - talloc_free(h); - return ret; - } - - /* tell ctdbd that we are finished with our local commit */ - ctdbd_control_local(messaging_ctdbd_connection(), CTDB_CONTROL_TRANS2_FINISHED, - h->ctx->db_id, CTDB_CTRL_FLAG_NOREPLY, - tdb_null, NULL, NULL, NULL); +done: h->ctx->transaction = NULL; talloc_free(h); return 0; @@ -1314,6 +1141,7 @@ struct db_context *db_open_ctdb(TALLOC_CTX *mem_ctx, struct db_context *result; struct db_ctdb_ctx *db_ctdb; char *db_path; + struct ctdbd_connection *conn; if (!lp_clustering()) { DEBUG(10, ("Clustering disabled -- no ctdb\n")); @@ -1335,13 +1163,15 @@ struct db_context *db_open_ctdb(TALLOC_CTX *mem_ctx, db_ctdb->transaction = NULL; db_ctdb->db = result; - if (!NT_STATUS_IS_OK(ctdbd_db_attach(messaging_ctdbd_connection(),name, &db_ctdb->db_id, tdb_flags))) { + conn = messaging_ctdbd_connection(); + + if (!NT_STATUS_IS_OK(ctdbd_db_attach(conn, name, &db_ctdb->db_id, tdb_flags))) { DEBUG(0, ("ctdbd_db_attach failed for %s\n", name)); TALLOC_FREE(result); return NULL; } - db_path = ctdbd_dbpath(messaging_ctdbd_connection(), db_ctdb, db_ctdb->db_id); + db_path = ctdbd_dbpath(conn, db_ctdb, db_ctdb->db_id); result->persistent = ((tdb_flags & TDB_CLEAR_IF_FIRST) == 0); @@ -1361,6 +1191,16 @@ struct db_context *db_open_ctdb(TALLOC_CTX *mem_ctx, } talloc_free(db_path); + if (result->persistent) { + db_ctdb->lock_ctx = g_lock_ctx_init(db_ctdb, + ctdb_conn_msg_ctx(conn)); + if (db_ctdb->lock_ctx == NULL) { + DEBUG(0, ("g_lock_ctx_init failed\n")); + TALLOC_FREE(result); + return NULL; + } + } + result->private_data = (void *)db_ctdb; result->fetch_locked = db_ctdb_fetch_locked; result->fetch = db_ctdb_fetch; -- cgit