A newer version of this documentation is available.

View Latest

Retrieving documents

You can retrieve documents by using either regular reads or replica reads.

Documents are retrieved by using the lcb_get() function, which requests the document from the cluster. To request a document from the server, initialize an lcb_get_cmd_t structure:

static void got_document(lcb_t instance, const void *cookie, lcb_error_t err,
   const lcb_get_resp_t *resp)
  if (err == LCB_SUCCESS) {
    printf("Received document: %.*s\n", (int)resp->v.v0.nbytes, resp->v.v0.bytes);
  } else {
    fprintf(stderr, "Couldn’t retrieve item: %s\n", lcb_strerror(instance, err));

static void getDoc() {
  // ...
  lcb_set_get_callback(instance, got_document);

  lcb_get_cmd_t cmd = { 0 };
  const lcb_get_cmd_t *cmdlist[] = { &cmd };
  const char *key = "a_simple_key";
  cmd.v.v0.key = key;
  cmd.v.v0.nkey = strlen(key);

  lcb_error_t err = lcb_get(instance, NULL, 1, &cmdlist);
  if (err == LCB_SUCCESS) {
  } else {
    fprintf(stderr, "Couldn’t schedule get operation: %s\n", lcb_strerror(instance, err));

The previous block of code shows how to request a document from the cluster. The key and nkey fields describe the buffer and length of the key, while the lcb_get() function actually schedules the request to the server. In order to receive the document you must install a callback that will be invoked with the document as one of its parameters.

Reading from replicas

In the event that a node fails, all the items hosted on that node will become temporarily unavailable until the node is failed over. The cluster may be configured to automatically fail over a node as early as 30 seconds after a failover is detected; however, some applications might need higher response times and availability in such cases.

The replica read functionality allows the application to read data from a replica. The data might be older in case the active copy of the data (on the node which failed) did not replicate in time to the replica. Issuing a replica read operation in the C SDK is similar to issuing a normal get operation. Just use the lcb_get_replica_cmd_t and lcb_get_replica APIs:

lcb_get_replica_cmd_t cmd = { 0 };
lcb_get_replica_cmd_t *cmdlist[] = { &cmd };
cmd.v.v0.key = "key";
cmd.v.v0.nkey = strlen("key");
cmd.v.v0.strategy = LCB_REPLICA_FIRST;
err = lcb_get_replica(instance, NULL, 1, &cmdlist);

The callback for the operation is the same as the one for lcb_get(). The strategy field instructs the library how to fetch from the replica. In this case the library requests a copy from each replica (in sequence) and invoke the callback with the first successful reply.

Getting from a replica might return stale data. Use replica reads only in instances where having any copy of the document (even a potentially stale one) is better than having no copy.