Merging upstream version 1.45.3+dfsg.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

9 files changed, 1662 insertions, 0 deletions

diff --git a/src/fluent-bit/plugins/in_forward/CMakeLists.txt b/src/fluent-bit/plugins/in_forward/CMakeLists.txt
new file mode 100644
index 000000000..ce4f62728
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/CMakeLists.txt
@@ -0,0 +1,7 @@
+set(src
+  fw.c
+  fw_conn.c
+  fw_prot.c
+  fw_config.c)
+
+FLB_PLUGIN(in_forward "${src}" "")
diff --git a/src/fluent-bit/plugins/in_forward/fw.c b/src/fluent-bit/plugins/in_forward/fw.c
new file mode 100644
index 000000000..b85b198c6
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw.c
@@ -0,0 +1,325 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#include <fluent-bit/flb_info.h>
+#include <fluent-bit/flb_input.h>
+#include <fluent-bit/flb_engine.h>
+#include <fluent-bit/flb_downstream.h>
+#include <fluent-bit/flb_input_plugin.h>
+#include <fluent-bit/flb_network.h>
+#include <msgpack.h>
+
+#ifdef FLB_HAVE_UNIX_SOCKET
+#include <sys/socket.h>
+#include <sys/un.h>
+#include <sys/stat.h>
+#endif
+
+#include "fw.h"
+#include "fw_conn.h"
+#include "fw_config.h"
+
+#ifdef FLB_HAVE_UNIX_SOCKET
+static int remove_existing_socket_file(char *socket_path)
+{
+    struct stat file_data;
+    int         result;
+
+    result = stat(socket_path, &file_data);
+
+    if (result == -1) {
+        if (errno == ENOENT) {
+            return 0;
+        }
+
+        flb_errno();
+
+        return -1;
+    }
+
+    if (S_ISSOCK(file_data.st_mode) == 0) {
+        return -2;
+    }
+
+    result = unlink(socket_path);
+
+    if (result != 0) {
+        return -3;
+    }
+
+    return 0;
+}
+
+static int fw_unix_create(struct flb_in_fw_config *ctx)
+{
+    int ret;
+
+    ret = remove_existing_socket_file(ctx->unix_path);
+
+    if (ret != 0) {
+        if (ret == -2) {
+            flb_plg_error(ctx->ins,
+                          "%s exists and it is not a unix socket. Aborting",
+                          ctx->unix_path);
+        }
+        else {
+            flb_plg_error(ctx->ins,
+                          "could not remove existing unix socket %s. Aborting",
+                          ctx->unix_path);
+        }
+
+        return -1;
+    }
+
+    ctx->downstream = flb_downstream_create(FLB_TRANSPORT_UNIX_STREAM,
+                                            ctx->ins->flags,
+                                            ctx->unix_path,
+                                            0,
+                                            ctx->ins->tls,
+                                            ctx->ins->config,
+                                            &ctx->ins->net_setup);
+
+    if (ctx->downstream == NULL) {
+        return -1;
+    }
+
+    if (ctx->unix_perm_str) {
+        if (chmod(ctx->unix_path, ctx->unix_perm)) {
+            flb_errno();
+
+            flb_plg_error(ctx->ins, "cannot set permission on '%s' to %04o",
+                          ctx->unix_path, ctx->unix_perm);
+
+            return -1;
+        }
+    }
+
+    return 0;
+}
+#endif
+
+/*
+ * For a server event, the collection event means a new client have arrived, we
+ * accept the connection and create a new FW instance which will wait for
+ * MessagePack records.
+ */
+static int in_fw_collect(struct flb_input_instance *ins,
+                         struct flb_config *config, void *in_context)
+{
+    struct flb_connection   *connection;
+    struct fw_conn          *conn;
+    struct flb_in_fw_config *ctx;
+
+    ctx = in_context;
+
+    connection = flb_downstream_conn_get(ctx->downstream);
+
+    if (connection == NULL) {
+        flb_plg_error(ctx->ins, "could not accept new connection");
+
+        return -1;
+    }
+
+    if (!config->is_ingestion_active) {
+        flb_downstream_conn_release(connection);
+
+        return -1;
+    }
+
+    flb_plg_trace(ins, "new TCP connection arrived FD=%i", connection->fd);
+
+    conn = fw_conn_add(connection, ctx);
+
+    if (!conn) {
+        return -1;
+    }
+
+    return 0;
+}
+
+/* Initialize plugin */
+static int in_fw_init(struct flb_input_instance *ins,
+                      struct flb_config *config, void *data)
+{
+    unsigned short int       port;
+    int                      ret;
+    struct flb_in_fw_config *ctx;
+
+    (void) data;
+
+    /* Allocate space for the configuration */
+    ctx = fw_config_init(ins);
+    if (!ctx) {
+        return -1;
+    }
+
+    ctx->coll_fd = -1;
+    ctx->ins = ins;
+    mk_list_init(&ctx->connections);
+
+    /* Set the context */
+    flb_input_set_context(ins, ctx);
+
+    /* Unix Socket mode */
+    if (ctx->unix_path) {
+#ifndef FLB_HAVE_UNIX_SOCKET
+        flb_plg_error(ctx->ins, "unix address is not supported %s:%s. Aborting",
+                      ctx->listen, ctx->tcp_port);
+        fw_config_destroy(ctx);
+        return -1;
+#else
+        ret = fw_unix_create(ctx);
+        if (ret != 0) {
+            flb_plg_error(ctx->ins, "could not listen on unix://%s",
+                          ctx->unix_path);
+            fw_config_destroy(ctx);
+            return -1;
+        }
+        flb_plg_info(ctx->ins, "listening on unix://%s", ctx->unix_path);
+#endif
+    }
+    else {
+        port = (unsigned short int) strtoul(ctx->tcp_port, NULL, 10);
+
+        ctx->downstream = flb_downstream_create(FLB_TRANSPORT_TCP,
+                                                ctx->ins->flags,
+                                                ctx->listen,
+                                                port,
+                                                ctx->ins->tls,
+                                                config,
+                                                &ctx->ins->net_setup);
+
+        if (ctx->downstream == NULL) {
+            flb_plg_error(ctx->ins,
+                          "could not initialize downstream on unix://%s. Aborting",
+                          ctx->listen);
+
+            fw_config_destroy(ctx);
+
+            return -1;
+        }
+
+        if (ctx->downstream != NULL) {
+            flb_plg_info(ctx->ins, "listening on %s:%s",
+                         ctx->listen, ctx->tcp_port);
+        }
+        else {
+            flb_plg_error(ctx->ins, "could not bind address %s:%s. Aborting",
+                          ctx->listen, ctx->tcp_port);
+
+            fw_config_destroy(ctx);
+
+            return -1;
+        }
+    }
+
+    flb_input_downstream_set(ctx->downstream, ctx->ins);
+
+    flb_net_socket_nonblocking(ctx->downstream->server_fd);
+
+    /* Collect upon data available on the standard input */
+    ret = flb_input_set_collector_socket(ins,
+                                         in_fw_collect,
+                                         ctx->downstream->server_fd,
+                                         config);
+    if (ret == -1) {
+        flb_plg_error(ctx->ins, "could not set server socket collector");
+        fw_config_destroy(ctx);
+        return -1;
+    }
+
+    ctx->coll_fd = ret;
+
+    return 0;
+}
+
+static void in_fw_pause(void *data, struct flb_config *config)
+{
+    struct flb_in_fw_config *ctx = data;
+
+    /*
+     * If the plugin is paused AND the ingestion not longer active,
+     * it means we are in a shutdown phase. This plugin can safetly
+     * close the socket server collector.
+     *
+     * This socket stop is a workaround since the server API will be
+     * refactored shortly.
+     */
+    if (config->is_ingestion_active == FLB_FALSE) {
+        fw_conn_del_all(ctx);
+    }
+}
+
+static int in_fw_exit(void *data, struct flb_config *config)
+{
+    (void) *config;
+    struct flb_in_fw_config *ctx = data;
+
+    if (!ctx) {
+        return 0;
+    }
+
+    fw_conn_del_all(ctx);
+    fw_config_destroy(ctx);
+    return 0;
+}
+
+/* Configuration properties map */
+static struct flb_config_map config_map[] = {
+   {
+    FLB_CONFIG_MAP_STR, "tag_prefix", NULL,
+    0, FLB_TRUE, offsetof(struct flb_in_fw_config, tag_prefix),
+    "Prefix incoming tag with the defined value."
+   },
+   {
+    FLB_CONFIG_MAP_STR, "unix_path", NULL,
+    0, FLB_TRUE, offsetof(struct flb_in_fw_config, unix_path),
+    "The path to unix socket to receive a Forward message."
+   },
+   {
+    FLB_CONFIG_MAP_STR, "unix_perm", (char *)NULL,
+    0, FLB_TRUE, offsetof(struct flb_in_fw_config, unix_perm_str),
+    "Set the permissions for the UNIX socket"
+   },
+   {
+    FLB_CONFIG_MAP_SIZE, "buffer_chunk_size", FLB_IN_FW_CHUNK_SIZE,
+    0, FLB_TRUE, offsetof(struct flb_in_fw_config, buffer_chunk_size),
+    "The buffer memory size used to receive a Forward message."
+   },
+   {
+    FLB_CONFIG_MAP_SIZE, "buffer_max_size", FLB_IN_FW_CHUNK_MAX_SIZE,
+    0, FLB_TRUE, offsetof(struct flb_in_fw_config, buffer_max_size),
+    "The maximum buffer memory size used to receive a Forward message."
+   },
+   {0}
+};
+
+/* Plugin reference */
+struct flb_input_plugin in_forward_plugin = {
+    .name         = "forward",
+    .description  = "Fluentd in-forward",
+    .cb_init      = in_fw_init,
+    .cb_pre_run   = NULL,
+    .cb_collect   = in_fw_collect,
+    .cb_flush_buf = NULL,
+    .cb_pause     = in_fw_pause,
+    .cb_exit      = in_fw_exit,
+    .config_map   = config_map,
+    .flags        = FLB_INPUT_NET_SERVER | FLB_IO_OPT_TLS
+};
diff --git a/src/fluent-bit/plugins/in_forward/fw.h b/src/fluent-bit/plugins/in_forward/fw.h
new file mode 100644
index 000000000..454f255b9
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw.h
@@ -0,0 +1,52 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#ifndef FLB_IN_FW_H
+#define FLB_IN_FW_H
+
+#include <msgpack.h>
+#include <fluent-bit/flb_input.h>
+#include <fluent-bit/flb_log_event_decoder.h>
+#include <fluent-bit/flb_log_event_encoder.h>
+
+struct flb_in_fw_config {
+    size_t buffer_max_size;         /* Max Buffer size             */
+    size_t buffer_chunk_size;       /* Chunk allocation size       */
+
+    /* Network */
+    char *listen;                   /* Listen interface            */
+    char *tcp_port;                 /* TCP Port                    */
+
+    flb_sds_t tag_prefix;           /* tag prefix                  */
+
+    /* Unix Socket */
+    char *unix_path;                /* Unix path for socket        */
+    unsigned int unix_perm;         /* Permission for socket       */
+    flb_sds_t unix_perm_str;        /* Permission (config map)     */
+
+    int coll_fd;
+    struct flb_downstream *downstream; /* Client manager          */
+    struct mk_list connections;     /* List of active connections */
+    struct flb_input_instance *ins; /* Input plugin instace       */
+
+    struct flb_log_event_decoder *log_decoder;
+    struct flb_log_event_encoder *log_encoder;
+};
+
+#endif
diff --git a/src/fluent-bit/plugins/in_forward/fw_config.c b/src/fluent-bit/plugins/in_forward/fw_config.c
new file mode 100644
index 000000000..7edbab0c7
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw_config.c
@@ -0,0 +1,120 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#include <stdlib.h>
+#include <fluent-bit/flb_utils.h>
+#include <fluent-bit/flb_downstream.h>
+#include <fluent-bit/flb_input_plugin.h>
+
+#include "fw.h"
+#include "fw_conn.h"
+#include "fw_config.h"
+
+struct flb_in_fw_config *fw_config_init(struct flb_input_instance *i_ins)
+{
+    char tmp[16];
+    int ret = -1;
+    const char *p;
+    struct flb_in_fw_config *config;
+
+    config = flb_calloc(1, sizeof(struct flb_in_fw_config));
+    if (!config) {
+        flb_errno();
+        return NULL;
+    }
+    config->coll_fd = -1;
+
+    config->log_encoder = flb_log_event_encoder_create(FLB_LOG_EVENT_FORMAT_DEFAULT);
+
+    if (config->log_encoder == NULL) {
+        flb_plg_error(i_ins, "could not initialize event encoder");
+        fw_config_destroy(config);
+
+        return NULL;
+    }
+
+    config->log_decoder = flb_log_event_decoder_create(NULL, 0);
+
+    if (config->log_decoder == NULL) {
+        flb_plg_error(i_ins, "could not initialize event decoder");
+        fw_config_destroy(config);
+
+        return NULL;
+    }
+
+    ret = flb_input_config_map_set(i_ins, (void *)config);
+    if (ret == -1) {
+        flb_plg_error(i_ins, "config map set error");
+        flb_free(config);
+        return NULL;
+    }
+
+    p = flb_input_get_property("unix_path", i_ins);
+    if (p == NULL) {
+        /* Listen interface (if not set, defaults to 0.0.0.0:24224) */
+        flb_input_net_default_listener("0.0.0.0", 24224, i_ins);
+        config->listen = i_ins->host.listen;
+        snprintf(tmp, sizeof(tmp) - 1, "%d", i_ins->host.port);
+        config->tcp_port = flb_strdup(tmp);
+    }
+    else {
+        /* Unix socket mode */
+        if (config->unix_perm_str) {
+            config->unix_perm = strtol(config->unix_perm_str, NULL, 8) & 07777;
+        }
+    }
+
+    if (!config->unix_path) {
+        flb_debug("[in_fw] Listen='%s' TCP_Port=%s",
+                  config->listen, config->tcp_port);
+    }
+    return config;
+}
+
+int fw_config_destroy(struct flb_in_fw_config *config)
+{
+    if (config->log_encoder != NULL) {
+        flb_log_event_encoder_destroy(config->log_encoder);
+    }
+
+    if (config->log_decoder != NULL) {
+        flb_log_event_decoder_destroy(config->log_decoder);
+    }
+
+    if (config->coll_fd != -1) {
+        flb_input_collector_delete(config->coll_fd, config->ins);
+
+        config->coll_fd = -1;
+    }
+
+    if (config->downstream != NULL) {
+        flb_downstream_destroy(config->downstream);
+    }
+
+    if (config->unix_path) {
+        unlink(config->unix_path);
+    }
+    else {
+        flb_free(config->tcp_port);
+    }
+
+    flb_free(config);
+
+    return 0;
+}
diff --git a/src/fluent-bit/plugins/in_forward/fw_config.h b/src/fluent-bit/plugins/in_forward/fw_config.h
new file mode 100644
index 000000000..bbad17610
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw_config.h
@@ -0,0 +1,28 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#ifndef FLB_IN_FW_CONFIG_H
+#define FLB_IN_FW_CONFIG_H
+
+#include "fw.h"
+
+struct flb_in_fw_config *fw_config_init(struct flb_input_instance *i_ins);
+int fw_config_destroy(struct flb_in_fw_config *config);
+
+#endif
diff --git a/src/fluent-bit/plugins/in_forward/fw_conn.c b/src/fluent-bit/plugins/in_forward/fw_conn.c
new file mode 100644
index 000000000..3ccd98c24
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw_conn.c
@@ -0,0 +1,199 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#include <fluent-bit/flb_info.h>
+#include <fluent-bit/flb_input_plugin.h>
+#include <fluent-bit/flb_utils.h>
+#include <fluent-bit/flb_engine.h>
+#include <fluent-bit/flb_network.h>
+#include <fluent-bit/flb_downstream.h>
+
+#include "fw.h"
+#include "fw_prot.h"
+#include "fw_conn.h"
+
+/* Callback invoked every time an event is triggered for a connection */
+int fw_conn_event(void *data)
+{
+    int ret;
+    int bytes;
+    int available;
+    int size;
+    char *tmp;
+    struct fw_conn *conn;
+    struct mk_event *event;
+    struct flb_in_fw_config *ctx;
+    struct flb_connection *connection;
+
+    connection = (struct flb_connection *) data;
+
+    conn = connection->user_data;
+
+    ctx = conn->ctx;
+
+    event = &connection->event;
+
+
+    if (event->mask & MK_EVENT_READ) {
+        available = (conn->buf_size - conn->buf_len);
+        if (available < 1) {
+            if (conn->buf_size >= ctx->buffer_max_size) {
+                flb_plg_warn(ctx->ins, "fd=%i incoming data exceed limit (%lu bytes)",
+                             event->fd, (ctx->buffer_max_size));
+                fw_conn_del(conn);
+                return -1;
+            }
+            else if (conn->buf_size + ctx->buffer_chunk_size > ctx->buffer_max_size) {
+                /* no space to add buffer_chunk_size */
+                /* set maximum size */
+                size = ctx->buffer_max_size;
+            }
+            else {
+                size = conn->buf_size + ctx->buffer_chunk_size;
+            }
+            tmp = flb_realloc(conn->buf, size);
+            if (!tmp) {
+                flb_errno();
+                return -1;
+            }
+            flb_plg_trace(ctx->ins, "fd=%i buffer realloc %i -> %i",
+                          event->fd, conn->buf_size, size);
+
+            conn->buf = tmp;
+            conn->buf_size = size;
+            available = (conn->buf_size - conn->buf_len);
+        }
+
+        bytes = flb_io_net_read(connection,
+                                (void *) &conn->buf[conn->buf_len],
+                                available);
+
+        if (bytes > 0) {
+            flb_plg_trace(ctx->ins, "read()=%i pre_len=%i now_len=%i",
+                          bytes, conn->buf_len, conn->buf_len + bytes);
+            conn->buf_len += bytes;
+
+            ret = fw_prot_process(ctx->ins, conn);
+            if (ret == -1) {
+                fw_conn_del(conn);
+                return -1;
+            }
+            return bytes;
+        }
+        else {
+            flb_plg_trace(ctx->ins, "fd=%i closed connection", event->fd);
+            fw_conn_del(conn);
+            return -1;
+        }
+    }
+
+    if (event->mask & MK_EVENT_CLOSE) {
+        flb_plg_trace(ctx->ins, "fd=%i hangup", event->fd);
+        fw_conn_del(conn);
+        return -1;
+    }
+    return 0;
+}
+
+/* Create a new Forward request instance */
+struct fw_conn *fw_conn_add(struct flb_connection *connection, struct flb_in_fw_config *ctx)
+{
+    struct fw_conn *conn;
+    int             ret;
+
+    conn = flb_malloc(sizeof(struct fw_conn));
+    if (!conn) {
+        flb_errno();
+
+        return NULL;
+    }
+
+    conn->connection = connection;
+
+    /* Set data for the event-loop */
+    connection->user_data     = conn;
+    connection->event.type    = FLB_ENGINE_EV_CUSTOM;
+    connection->event.handler = fw_conn_event;
+
+    /* Connection info */
+    conn->ctx     = ctx;
+    conn->buf_len = 0;
+    conn->rest    = 0;
+    conn->status  = FW_NEW;
+
+    /* Allocate read buffer */
+    conn->buf = flb_malloc(ctx->buffer_chunk_size);
+    if (!conn->buf) {
+        flb_errno();
+        flb_free(conn);
+
+        return NULL;
+    }
+    conn->buf_size = ctx->buffer_chunk_size;
+    conn->in       = ctx->ins;
+
+    /* Register instance into the event loop */
+    ret = mk_event_add(flb_engine_evl_get(),
+                       connection->fd,
+                       FLB_ENGINE_EV_CUSTOM,
+                       MK_EVENT_READ,
+                       &connection->event);
+    if (ret == -1) {
+        flb_plg_error(ctx->ins, "could not register new connection");
+
+        flb_free(conn->buf);
+        flb_free(conn);
+
+        return NULL;
+    }
+
+    mk_list_add(&conn->_head, &ctx->connections);
+
+    return conn;
+}
+
+int fw_conn_del(struct fw_conn *conn)
+{
+    /* The downstream unregisters the file descriptor from the event-loop
+     * so there's nothing to be done by the plugin
+     */
+    flb_downstream_conn_release(conn->connection);
+
+    /* Release resources */
+    mk_list_del(&conn->_head);
+
+    flb_free(conn->buf);
+    flb_free(conn);
+
+    return 0;
+}
+
+int fw_conn_del_all(struct flb_in_fw_config *ctx)
+{
+    struct mk_list *tmp;
+    struct mk_list *head;
+    struct fw_conn *conn;
+
+    mk_list_foreach_safe(head, tmp, &ctx->connections) {
+        conn = mk_list_entry(head, struct fw_conn, _head);
+        fw_conn_del(conn);
+    }
+
+    return 0;
+}
\ No newline at end of file
diff --git a/src/fluent-bit/plugins/in_forward/fw_conn.h b/src/fluent-bit/plugins/in_forward/fw_conn.h
new file mode 100644
index 000000000..4c04d9400
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw_conn.h
@@ -0,0 +1,57 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#ifndef FLB_IN_FW_CONN_H
+#define FLB_IN_FW_CONN_H
+
+#define FLB_IN_FW_CHUNK_SIZE      "1024000" /* 1MB */
+#define FLB_IN_FW_CHUNK_MAX_SIZE  "6144000" /* =FLB_IN_FW_CHUNK_SIZE * 6.  6MB */
+
+enum {
+    FW_NEW        = 1,  /* it's a new connection                */
+    FW_CONNECTED  = 2,  /* MQTT connection per protocol spec OK */
+};
+
+struct fw_conn_stream {
+    char *tag;
+    size_t tag_len;
+};
+
+/* Respresents a connection */
+struct fw_conn {
+    int status;                      /* Connection status                 */
+
+    /* Buffer */
+    char *buf;                       /* Buffer data                       */
+    int  buf_len;                    /* Data length                       */
+    int  buf_size;                   /* Buffer size                       */
+    size_t rest;                     /* Unpacking offset                  */
+
+    struct flb_input_instance *in;   /* Parent plugin instance            */
+    struct flb_in_fw_config *ctx;    /* Plugin configuration context      */
+    struct flb_connection *connection;
+
+    struct mk_list _head;
+};
+
+struct fw_conn *fw_conn_add(struct flb_connection *connection, struct flb_in_fw_config *ctx);
+int fw_conn_del(struct fw_conn *conn);
+int fw_conn_del_all(struct flb_in_fw_config *ctx);
+
+#endif
diff --git a/src/fluent-bit/plugins/in_forward/fw_prot.c b/src/fluent-bit/plugins/in_forward/fw_prot.c
new file mode 100644
index 000000000..2a23b6254
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw_prot.c
@@ -0,0 +1,846 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#include <fluent-bit/flb_info.h>
+#include <fluent-bit/flb_input_plugin.h>
+#include <fluent-bit/flb_input.h>
+#include <fluent-bit/flb_config.h>
+#include <fluent-bit/flb_pack.h>
+#include <fluent-bit/flb_utils.h>
+#include <fluent-bit/flb_gzip.h>
+
+#include <fluent-bit/flb_input_metric.h>
+#include <fluent-bit/flb_input_trace.h>
+
+#include <cmetrics/cmetrics.h>
+#include <cmetrics/cmt_decode_msgpack.h>
+
+#include <ctraces/ctraces.h>
+#include <ctraces/ctr_decode_msgpack.h>
+
+#include <msgpack.h>
+
+#include "fw.h"
+#include "fw_prot.h"
+#include "fw_conn.h"
+
+/* Try parsing rounds up-to 32 bytes */
+#define EACH_RECV_SIZE 32
+
+static int get_chunk_event_type(struct flb_input_instance *ins, msgpack_object options)
+{
+    int i;
+    int type = FLB_EVENT_TYPE_LOGS;
+    msgpack_object k;
+    msgpack_object v;
+
+    if (options.type != MSGPACK_OBJECT_MAP) {
+        flb_plg_error(ins, "invalid options field in record");
+        return -1;
+    }
+
+    for (i = 0; i < options.via.map.size; i++) {
+        k = options.via.map.ptr[i].key;
+        v = options.via.map.ptr[i].val;
+
+        if (k.type != MSGPACK_OBJECT_STR) {
+            return -1;
+        }
+
+        if (k.via.str.size != 13) {
+            continue;
+        }
+
+        if (strncmp(k.via.str.ptr, "fluent_signal", 13) == 0) {
+            if (v.type != MSGPACK_OBJECT_POSITIVE_INTEGER) {
+                flb_plg_error(ins, "invalid value type in options fluent_signal");
+                return -1;
+            }
+
+            if (v.via.i64 != FLB_EVENT_TYPE_LOGS && v.via.i64 != FLB_EVENT_TYPE_METRICS && v.via.i64 != FLB_EVENT_TYPE_TRACES) {
+                flb_plg_error(ins, "invalid value in options fluent_signal");
+                return -1;
+            }
+
+            /* cast should be fine */
+            type = (int) v.via.i64;
+            break;
+        }
+    }
+
+    return type;
+}
+
+static int is_gzip_compressed(msgpack_object options)
+{
+    int i;
+    msgpack_object k;
+    msgpack_object v;
+
+    if (options.type != MSGPACK_OBJECT_MAP) {
+        return -1;
+    }
+
+
+    for (i = 0; i < options.via.map.size; i++) {
+        k = options.via.map.ptr[i].key;
+        v = options.via.map.ptr[i].val;
+
+        if (k.type != MSGPACK_OBJECT_STR) {
+            return -1;
+        }
+
+        if (k.via.str.size != 10) {
+            continue;
+        }
+
+        if (strncmp(k.via.str.ptr, "compressed", 10) == 0) {
+            if (v.type != MSGPACK_OBJECT_STR) {
+                return -1;
+            }
+
+            if (v.via.str.size != 4) {
+                return -1;
+            }
+
+            if (strncmp(v.via.str.ptr, "gzip", 4) == 0) {
+                return FLB_TRUE;
+            }
+            else if (strncmp(v.via.str.ptr, "text", 4) == 0) {
+                return FLB_FALSE;
+            }
+
+            return -1;
+        }
+    }
+
+    return FLB_FALSE;
+}
+
+static int send_ack(struct flb_input_instance *in, struct fw_conn *conn,
+                    msgpack_object chunk)
+{
+    int result;
+    size_t sent;
+    ssize_t bytes;
+    msgpack_packer mp_pck;
+    msgpack_sbuffer mp_sbuf;
+
+    msgpack_sbuffer_init(&mp_sbuf);
+    msgpack_packer_init(&mp_pck, &mp_sbuf, msgpack_sbuffer_write);
+
+    msgpack_pack_map(&mp_pck, 1);
+    msgpack_pack_str(&mp_pck, 3);
+    msgpack_pack_str_body(&mp_pck, "ack", 3);
+    msgpack_pack_object(&mp_pck, chunk);
+
+
+    bytes = flb_io_net_write(conn->connection,
+                             (void *) mp_sbuf.data,
+                             mp_sbuf.size,
+                             &sent);
+
+    msgpack_sbuffer_destroy(&mp_sbuf);
+
+    if (bytes == -1) {
+        flb_plg_error(in, "cannot send ACK response: %.*s",
+                      chunk.via.str.size, chunk.via.str.ptr);
+
+        result = -1;
+    }
+    else {
+        result = 0;
+    }
+
+    return result;
+
+}
+
+static size_t get_options_metadata(msgpack_object *arr, int expected, size_t *idx)
+{
+    size_t i;
+    msgpack_object *options;
+    msgpack_object k;
+    msgpack_object v;
+
+    if (arr->type != MSGPACK_OBJECT_ARRAY) {
+        return -1;
+    }
+
+    /* Make sure the 'expected' entry position is valid for the array size */
+    if (expected >= arr->via.array.size) {
+        return 0;
+    }
+
+    options = &arr->via.array.ptr[expected];
+    if (options->type == MSGPACK_OBJECT_NIL) {
+        /*
+         * Old Docker 18.x sends a NULL options parameter, just be friendly and
+         * let it pass.
+         */
+        return 0;
+    }
+
+    if (options->type != MSGPACK_OBJECT_MAP) {
+        return -1;
+    }
+
+    if (options->via.map.size <= 0) {
+        return 0;
+    }
+
+    for (i = 0; i < options->via.map.size; i++) {
+        k = options->via.map.ptr[i].key;
+        v = options->via.map.ptr[i].val;
+
+        if (k.type != MSGPACK_OBJECT_STR) {
+            continue;
+        }
+
+        if (k.via.str.size != 8) {
+            continue;
+        }
+
+        if (strncmp(k.via.str.ptr, "metadata", 8) != 0) {
+            continue;
+        }
+
+        if (v.type != MSGPACK_OBJECT_MAP) {
+            return -1;
+        }
+
+        *idx = i;
+
+        return 0;
+    }
+
+    return 0;
+}
+
+static size_t get_options_chunk(msgpack_object *arr, int expected, size_t *idx)
+{
+    size_t i;
+    msgpack_object *options;
+    msgpack_object k;
+    msgpack_object v;
+
+    if (arr->type != MSGPACK_OBJECT_ARRAY) {
+        return -1;
+    }
+
+    /* Make sure the 'expected' entry position is valid for the array size */
+    if (expected >= arr->via.array.size) {
+        return 0;
+    }
+
+    options = &arr->via.array.ptr[expected];
+    if (options->type == MSGPACK_OBJECT_NIL) {
+        /*
+         * Old Docker 18.x sends a NULL options parameter, just be friendly and
+         * let it pass.
+         */
+        return 0;
+    }
+
+    if (options->type != MSGPACK_OBJECT_MAP) {
+        return -1;
+    }
+
+    if (options->via.map.size <= 0) {
+        return 0;
+    }
+
+    for (i = 0; i < options->via.map.size; i++) {
+        k = options->via.map.ptr[i].key;
+        v = options->via.map.ptr[i].val;
+
+        if (k.type != MSGPACK_OBJECT_STR) {
+            continue;
+        }
+
+        if (k.via.str.size != 5) {
+            continue;
+        }
+
+        if (strncmp(k.via.str.ptr, "chunk", 5) != 0) {
+            continue;
+        }
+
+        if (v.type != MSGPACK_OBJECT_STR) {
+            return -1;
+        }
+
+        *idx = i;
+        return 0;
+    }
+
+    return 0;
+}
+
+static int fw_process_forward_mode_entry(
+                struct fw_conn *conn,
+                const char *tag, int tag_len,
+                msgpack_object *entry,
+                int chunk_id)
+{
+    int                  result;
+    struct flb_log_event event;
+
+    result = flb_event_decoder_decode_object(conn->ctx->log_decoder,
+                                             &event, entry);
+
+    if (result == FLB_EVENT_DECODER_SUCCESS) {
+        result = flb_log_event_encoder_begin_record(conn->ctx->log_encoder);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_set_timestamp(conn->ctx->log_encoder,
+                                                     &event.timestamp);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_set_metadata_from_msgpack_object(
+                    conn->ctx->log_encoder,
+                    event.metadata);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_set_body_from_msgpack_object(
+                    conn->ctx->log_encoder,
+                    event.body);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_commit_record(conn->ctx->log_encoder);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        flb_input_log_append(conn->ctx->ins, tag, tag_len,
+                             conn->ctx->log_encoder->output_buffer,
+                             conn->ctx->log_encoder->output_length);
+    }
+
+    flb_log_event_encoder_reset(conn->ctx->log_encoder);
+
+    if (result != FLB_EVENT_ENCODER_SUCCESS) {
+        flb_plg_warn(conn->ctx->ins, "Event decoder failure : %d", result);
+
+        return -1;
+    }
+
+    return 0;
+}
+
+static int fw_process_message_mode_entry(
+                struct flb_input_instance *in,
+                struct fw_conn *conn,
+                const char *tag, int tag_len,
+                msgpack_object *root,
+                msgpack_object *ts,
+                msgpack_object *body,
+                int chunk_id, int metadata_id)
+{
+    struct flb_time  timestamp;
+    msgpack_object  *metadata;
+    msgpack_object   options;
+    int              result;
+    msgpack_object   chunk;
+
+    metadata = NULL;
+
+    if (chunk_id != -1 || metadata_id != -1) {
+        options = root->via.array.ptr[3];
+
+        if (metadata_id != -1) {
+            metadata = &options.via.map.ptr[metadata_id].val;
+        }
+    }
+
+    result = flb_log_event_decoder_decode_timestamp(ts, &timestamp);
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_begin_record(conn->ctx->log_encoder);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_set_timestamp(conn->ctx->log_encoder,
+                                                     &timestamp);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        if (metadata != NULL) {
+            result = flb_log_event_encoder_set_metadata_from_msgpack_object(
+                        conn->ctx->log_encoder,
+                        metadata);
+        }
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_set_body_from_msgpack_object(
+                    conn->ctx->log_encoder,
+                    body);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        result = flb_log_event_encoder_commit_record(conn->ctx->log_encoder);
+    }
+
+    if (result == FLB_EVENT_ENCODER_SUCCESS) {
+        flb_input_log_append(in, tag, tag_len,
+                             conn->ctx->log_encoder->output_buffer,
+                             conn->ctx->log_encoder->output_length);
+    }
+
+    flb_log_event_encoder_reset(conn->ctx->log_encoder);
+
+    if (chunk_id != -1) {
+        chunk = options.via.map.ptr[chunk_id].val;
+        send_ack(in, conn, chunk);
+    }
+
+    return 0;
+}
+
+static size_t receiver_recv(struct fw_conn *conn, char *buf, size_t try_size) {
+    size_t off;
+    size_t actual_size;
+
+    off = conn->buf_len - conn->rest;
+    actual_size = try_size;
+
+    if (actual_size > conn->rest) {
+        actual_size = conn->rest;
+    }
+
+    memcpy(buf, conn->buf + off, actual_size);
+    conn->rest -= actual_size;
+
+    return actual_size;
+}
+
+static size_t receiver_to_unpacker(struct fw_conn *conn, size_t request_size,
+                                   msgpack_unpacker *unpacker)
+{
+    size_t recv_len;
+
+    /* make sure there's enough room, or expand the unpacker accordingly */
+    if (msgpack_unpacker_buffer_capacity(unpacker) < request_size) {
+        msgpack_unpacker_reserve_buffer(unpacker, request_size);
+        assert(msgpack_unpacker_buffer_capacity(unpacker) >= request_size);
+    }
+    recv_len = receiver_recv(conn, msgpack_unpacker_buffer(unpacker),
+                             request_size);
+    msgpack_unpacker_buffer_consumed(unpacker, recv_len);
+
+    return recv_len;
+}
+
+int fw_prot_process(struct flb_input_instance *ins, struct fw_conn *conn)
+{
+    int ret;
+    int stag_len;
+    int event_type;
+    int contain_options = FLB_FALSE;
+    size_t index = 0;
+    size_t off = 0;
+    size_t chunk_id = -1;
+    size_t metadata_id = -1;
+    const char *stag;
+    flb_sds_t out_tag = NULL;
+    size_t bytes;
+    size_t recv_len;
+    size_t gz_size;
+    void *gz_data;
+    msgpack_object tag;
+    msgpack_object entry;
+    msgpack_object map;
+    msgpack_object root;
+    msgpack_object chunk;
+    msgpack_unpacked result;
+    msgpack_unpacker *unp;
+    size_t all_used = 0;
+    struct flb_in_fw_config *ctx = conn->ctx;
+    struct cmt *cmt;
+    struct ctrace *ctr;
+
+    /*
+     * [tag, time, record]
+     * [tag, [[time,record], [time,record], ...]]
+     */
+
+    out_tag = flb_sds_create_size(1024);
+    if (!out_tag) {
+        return -1;
+    }
+
+    unp = msgpack_unpacker_new(1024);
+    msgpack_unpacked_init(&result);
+    conn->rest = conn->buf_len;
+
+    while (1) {
+        recv_len = receiver_to_unpacker(conn, EACH_RECV_SIZE, unp);
+        if (recv_len == 0) {
+            /* No more data */
+            msgpack_unpacker_free(unp);
+            msgpack_unpacked_destroy(&result);
+
+            /* Adjust buffer data */
+            if (conn->buf_len >= all_used && all_used > 0) {
+                memmove(conn->buf, conn->buf + all_used,
+                        conn->buf_len - all_used);
+                conn->buf_len -= all_used;
+            }
+            flb_sds_destroy(out_tag);
+
+            return 0;
+        }
+
+        /* Always summarize the total number of bytes requested to parse */
+        ret = msgpack_unpacker_next_with_size(unp, &result, &bytes);
+
+        /*
+         * Upon parsing or memory errors, break the loop, return the error
+         * and expect the connection to be closed.
+         */
+        if (ret == MSGPACK_UNPACK_PARSE_ERROR ||
+            ret == MSGPACK_UNPACK_NOMEM_ERROR) {
+            /* A bit redunant, print out the real error */
+            if (ret == MSGPACK_UNPACK_PARSE_ERROR) {
+                flb_plg_debug(ctx->ins, "err=MSGPACK_UNPACK_PARSE_ERROR");
+            }
+            else {
+                flb_plg_error(ctx->ins, "err=MSGPACK_UNPACK_NOMEM_ERROR");
+            }
+
+            /* Cleanup buffers */
+            msgpack_unpacked_destroy(&result);
+            msgpack_unpacker_free(unp);
+            flb_sds_destroy(out_tag);
+
+            return -1;
+        }
+
+        while (ret == MSGPACK_UNPACK_SUCCESS) {
+            /*
+             * For buffering optimization we always want to know the total
+             * number of bytes involved on the new object returned. Despites
+             * buf_off always know the given bytes, it's likely we used a bit
+             * less. This 'all_used' field keep a reference per object so
+             * when returning to the caller we can adjust the source buffer
+             * and deprecated consumed data.
+             *
+             * The 'last_parsed' field is Fluent Bit specific and is documented
+             * in:
+             *
+             *  lib/msgpack-c/include/msgpack/unpack.h
+             *
+             * Other references:
+             *
+             *  https://github.com/msgpack/msgpack-c/issues/514
+             */
+            all_used += bytes;
+
+
+            /* Map the array */
+            root = result.data;
+
+            if (root.type != MSGPACK_OBJECT_ARRAY) {
+                flb_plg_debug(ctx->ins,
+                              "parser: expecting an array (type=%i), skip.",
+                              root.type);
+                msgpack_unpacked_destroy(&result);
+                msgpack_unpacker_free(unp);
+                flb_sds_destroy(out_tag);
+
+                return -1;
+            }
+
+            if (root.via.array.size < 2) {
+                flb_plg_debug(ctx->ins,
+                              "parser: array of invalid size, skip.");
+                msgpack_unpacked_destroy(&result);
+                msgpack_unpacker_free(unp);
+                flb_sds_destroy(out_tag);
+
+                return -1;
+            }
+
+            if (root.via.array.size == 3) {
+                contain_options = FLB_TRUE;
+            }
+
+            /* Get the tag */
+            tag = root.via.array.ptr[0];
+            if (tag.type != MSGPACK_OBJECT_STR) {
+                flb_plg_debug(ctx->ins,
+                              "parser: invalid tag format, skip.");
+                msgpack_unpacked_destroy(&result);
+                msgpack_unpacker_free(unp);
+                flb_sds_destroy(out_tag);
+                return -1;
+            }
+
+            /* reference the tag associated with the record */
+            stag     = tag.via.str.ptr;
+            stag_len = tag.via.str.size;
+
+            /* clear out_tag before using */
+            flb_sds_len_set(out_tag, 0);
+
+            /* Prefix the incoming record tag with a custom prefix */
+            if (ctx->tag_prefix) {
+                /* prefix */
+                flb_sds_cat_safe(&out_tag,
+                                 ctx->tag_prefix, flb_sds_len(ctx->tag_prefix));
+                /* record tag */
+                flb_sds_cat_safe(&out_tag, stag, stag_len);
+            }
+            else if (ins->tag && !ins->tag_default) {
+                /* if the input plugin instance Tag has been manually set, use it */
+                flb_sds_cat_safe(&out_tag, ins->tag, flb_sds_len(ins->tag));
+            }
+            else {
+                /* use the tag from the record */
+                flb_sds_cat_safe(&out_tag, stag, stag_len);
+            }
+
+            entry = root.via.array.ptr[1];
+
+            if (entry.type == MSGPACK_OBJECT_ARRAY) {
+                /*
+                 * Forward format 1 (forward mode: [tag, [[time, map], ...]]
+                 */
+
+                /* Check for options */
+                chunk_id = -1;
+                ret = get_options_chunk(&root, 2, &chunk_id);
+                if (ret == -1) {
+                    flb_plg_debug(ctx->ins, "invalid options field");
+                    msgpack_unpacked_destroy(&result);
+                    msgpack_unpacker_free(unp);
+                    flb_sds_destroy(out_tag);
+
+                    return -1;
+                }
+
+                /* Process array */
+                ret = 0;
+
+                for(index = 0 ;
+                    index < entry.via.array.size &&
+                    ret == 0 ;
+                    index++) {
+                    ret = fw_process_forward_mode_entry(
+                            conn,
+                            out_tag, flb_sds_len(out_tag),
+                            &entry.via.array.ptr[index],
+                            chunk_id);
+                }
+
+                if (chunk_id != -1) {
+                    msgpack_object options;
+                    msgpack_object chunk;
+
+                    options = root.via.array.ptr[2];
+                    chunk = options.via.map.ptr[chunk_id].val;
+
+                    send_ack(conn->in, conn, chunk);
+                }
+            }
+            else if (entry.type == MSGPACK_OBJECT_POSITIVE_INTEGER ||
+                     entry.type == MSGPACK_OBJECT_EXT) {
+                /*
+                 * Forward format 2 (message mode) : [tag, time, map, ...]
+                 */
+                map = root.via.array.ptr[2];
+                if (map.type != MSGPACK_OBJECT_MAP) {
+                    flb_plg_warn(ctx->ins, "invalid data format, map expected");
+                    msgpack_unpacked_destroy(&result);
+                    msgpack_unpacker_free(unp);
+                    flb_sds_destroy(out_tag);
+                    return -1;
+                }
+
+                /* Check for options */
+                chunk_id = -1;
+                ret = get_options_chunk(&root, 3, &chunk_id);
+                if (ret == -1) {
+                    flb_plg_debug(ctx->ins, "invalid options field");
+                    msgpack_unpacked_destroy(&result);
+                    msgpack_unpacker_free(unp);
+                    flb_sds_destroy(out_tag);
+                    return -1;
+                }
+
+                metadata_id = -1;
+                ret = get_options_metadata(&root, 3, &metadata_id);
+                if (ret == -1) {
+                    flb_plg_debug(ctx->ins, "invalid options field");
+                    msgpack_unpacked_destroy(&result);
+                    msgpack_unpacker_free(unp);
+                    flb_sds_destroy(out_tag);
+                    return -1;
+                }
+
+                /* Process map */
+                fw_process_message_mode_entry(
+                    conn->in, conn,
+                    out_tag, flb_sds_len(out_tag),
+                    &root, &entry, &map, chunk_id,
+                    metadata_id);
+            }
+            else if (entry.type == MSGPACK_OBJECT_STR ||
+                     entry.type == MSGPACK_OBJECT_BIN) {
+                /* PackedForward Mode */
+                const char *data = NULL;
+                size_t len = 0;
+
+                /* Check for options */
+                chunk_id = -1;
+                ret = get_options_chunk(&root, 2, &chunk_id);
+                if (ret == -1) {
+                    flb_plg_debug(ctx->ins, "invalid options field");
+                    msgpack_unpacked_destroy(&result);
+                    msgpack_unpacker_free(unp);
+                    flb_sds_destroy(out_tag);
+                    return -1;
+                }
+
+                if (entry.type == MSGPACK_OBJECT_STR) {
+                    data = entry.via.str.ptr;
+                    len = entry.via.str.size;
+                }
+                else if (entry.type == MSGPACK_OBJECT_BIN) {
+                    data = entry.via.bin.ptr;
+                    len = entry.via.bin.size;
+                }
+
+                if (data) {
+                    ret = is_gzip_compressed(root.via.array.ptr[2]);
+                    if (ret == -1) {
+                        flb_plg_error(ctx->ins, "invalid 'compressed' option");
+                        msgpack_unpacked_destroy(&result);
+                        msgpack_unpacker_free(unp);
+                        flb_sds_destroy(out_tag);
+                        return -1;
+                    }
+
+                    if (ret == FLB_TRUE) {
+                        ret = flb_gzip_uncompress((void *) data, len,
+                                                  &gz_data, &gz_size);
+                        if (ret == -1) {
+                            flb_plg_error(ctx->ins, "gzip uncompress failure");
+                            msgpack_unpacked_destroy(&result);
+                            msgpack_unpacker_free(unp);
+                            flb_sds_destroy(out_tag);
+                            return -1;
+                        }
+
+                        /* Append uncompressed data */
+                        flb_input_log_append(conn->in,
+                                             out_tag, flb_sds_len(out_tag),
+                                             gz_data, gz_size);
+                        flb_free(gz_data);
+                    }
+                    else {
+                        event_type = FLB_EVENT_TYPE_LOGS;
+                        if (contain_options) {
+                            ret = get_chunk_event_type(ins, root.via.array.ptr[2]);
+                            if (ret == -1) {
+                                msgpack_unpacked_destroy(&result);
+                                msgpack_unpacker_free(unp);
+                                flb_sds_destroy(out_tag);
+                                return -1;
+                            }
+                            event_type = ret;
+                        }
+
+                        if (event_type == FLB_EVENT_TYPE_LOGS) {
+                            flb_input_log_append(conn->in,
+                                                 out_tag, flb_sds_len(out_tag),
+                                                 data, len);
+                        }
+                        else if (event_type == FLB_EVENT_TYPE_METRICS) {
+                            ret = cmt_decode_msgpack_create(&cmt, (char *) data, len, &off);
+                            if (ret == -1) {
+                                msgpack_unpacked_destroy(&result);
+                                msgpack_unpacker_free(unp);
+                                flb_sds_destroy(out_tag);
+                                return -1;
+                            }
+                            flb_input_metrics_append(conn->in,
+                                                     out_tag, flb_sds_len(out_tag),
+                                                     cmt);
+                        }
+                        else if (event_type == FLB_EVENT_TYPE_TRACES) {
+                            off = 0;
+                            ret = ctr_decode_msgpack_create(&ctr, (char *) data, len, &off);
+                            if (ret == -1) {
+                                msgpack_unpacked_destroy(&result);
+                                msgpack_unpacker_free(unp);
+                                flb_sds_destroy(out_tag);
+                                return -1;
+                            }
+
+                            flb_input_trace_append(ins,
+                                                   out_tag, flb_sds_len(out_tag),
+                                                   ctr);
+                        }
+                    }
+
+                    /* Handle ACK response */
+                    if (chunk_id != -1) {
+                        chunk = root.via.array.ptr[2].via.map.ptr[chunk_id].val;
+                        send_ack(ctx->ins, conn, chunk);
+                    }
+                }
+            }
+            else {
+                flb_plg_warn(ctx->ins, "invalid data format, type=%i",
+                             entry.type);
+                msgpack_unpacked_destroy(&result);
+                msgpack_unpacker_free(unp);
+                return -1;
+            }
+
+            ret = msgpack_unpacker_next(unp, &result);
+        }
+    }
+
+    msgpack_unpacked_destroy(&result);
+    msgpack_unpacker_free(unp);
+    flb_sds_destroy(out_tag);
+
+    switch (ret) {
+    case MSGPACK_UNPACK_EXTRA_BYTES:
+        flb_plg_error(ctx->ins, "MSGPACK_UNPACK_EXTRA_BYTES");
+        return -1;
+    case MSGPACK_UNPACK_CONTINUE:
+        flb_plg_trace(ctx->ins, "MSGPACK_UNPACK_CONTINUE");
+        return 1;
+    case MSGPACK_UNPACK_PARSE_ERROR:
+        flb_plg_debug(ctx->ins, "err=MSGPACK_UNPACK_PARSE_ERROR");
+        return -1;
+    case MSGPACK_UNPACK_NOMEM_ERROR:
+        flb_plg_error(ctx->ins, "err=MSGPACK_UNPACK_NOMEM_ERROR");
+        return -1;
+    };
+
+    return 0;
+}
diff --git a/src/fluent-bit/plugins/in_forward/fw_prot.h b/src/fluent-bit/plugins/in_forward/fw_prot.h
new file mode 100644
index 000000000..67eae5507
--- /dev/null
+++ b/src/fluent-bit/plugins/in_forward/fw_prot.h
@@ -0,0 +1,28 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+
+/*  Fluent Bit
+ *  ==========
+ *  Copyright (C) 2015-2022 The Fluent Bit Authors
+ *
+ *  Licensed under the Apache License, Version 2.0 (the "License");
+ *  you may not use this file except in compliance with the License.
+ *  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ *  Unless required by applicable law or agreed to in writing, software
+ *  distributed under the License is distributed on an "AS IS" BASIS,
+ *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ *  See the License for the specific language governing permissions and
+ *  limitations under the License.
+ */
+
+#ifndef FLB_IN_FW_PROT_H
+#define FLB_IN_FW_PROT_H
+
+#include "fw_conn.h"
+
+int fw_prot_parser(struct fw_conn *conn);
+int fw_prot_process(struct flb_input_instance *ins, struct fw_conn *conn);
+
+#endif