See 0.12.x branch for Fluentd 0.12.x.
Fluentd input plugin that acts as sFlow v2/v4/v5 collector.
Including a ruby wrapper of sflowtool which processes sFlow datagrams to report in JSON format.
fluent-plugin-sflow supports many packet formats below. Basically, it can process anything sflowtool can do. :sparkles:
See sflowtool document for more details.
gcc
and make
You need to install development tools beforehand.
:memo: If you see any problem on other platforms rather than Debian, Ubuntu, or CentOS, please open a new issue.
$ sudo apt install -y gcc make
If you're using ruby installed by apt
, install ruby-dev
.
( Skip if your ruby
is installed by rbenv
)
$ sudo apt install -y ruby-dev
$ sudo dnf install -y redhat-rpm-config gcc make
If you're using ruby installed by dnf
( or yum
), install ruby-devel
.
( Skip if your ruby
is installed by rbenv
)
$ sudo dnf install -y ruby-devel
$ gem install fluent-plugin-sflow
You can also use fluent-gem
or td-agent-gem
instead of gem
command.
<source>
@type sflow
tag example.sflow
bind 0.0.0.0
port 6343
</source>
bind
IP address on which this plugin will accept sFlow. (Default: '0.0.0.0')
port
UDP port number on which this plugin will accept sFlow. (Default: 6343)
{
"datagram_source_ip": "10.1.2.1",
"datagram_size": 240,
"unix_seconds_utc": 1502113217,
"datagram_version": 5,
"agent_sub_id": 0,
"agent": "10.1.2.1",
"packet_sequence_no": 53,
"sys_up_time": 3808877,
"samples_in_packet": 1,
"sample_type_tag": "0:1",
"sample_type": "flow_sample",
"sample_sequence_no": 44,
"source_id": "0:513",
"mean_skip_count": 5,
"sample_pool": 225,
"drop_events": 0,
"input_port": 513,
"output_port": 512,
"flow_block_tag": "0:1002",
"header_protocol": 1,
"sampled_packet_size": 102,
"stripped_bytes": 4,
"header_len": 98,
"header_bytes": "02-05-86-71-7D-03-08-00-27-12-99-09-08-00-45-00-00-54-B0-61-40-00-40-01-79-44-0A-01-02-02-01-00-04-01-08-00-DF-97-04-12-03-AD-D2-1A-87-59-00-00-00-00-F6-61-02-00-00-00-00-00-10-11-12-13-14-15-16-17-18-19-1A-1B-1C-1D-1E-1F-20-21-22-23-24-25-26-27-28-29-2A-2B-2C-2D-2E-2F-30-31-32-33-34-35-36-37",
"dst_mac": "020586717d03",
"src_mac": "080027129909",
"ip_size": 84,
"ip_tot_len": 84,
"src_ip": "10.1.2.2",
"dst_ip": "1.0.4.1",
"ip_protocol": 1,
"ip_tos": 0,
"ip_ttl": 64,
"ip_id": 25008,
"icmp_type": 8,
"icmp_code": 0,
"in_vlan": 0,
"in_priority": 0,
"out_vlan": 0,
"out_priority": 0,
"next_hop": "202.249.2.169",
"src_subnet_mask": 32,
"dst_subnet_mask": 22
}
{
"datagram_source_ip": "10.1.2.1",
"datagram_size": 204,
"unix_seconds_utc": 1502113415,
"datagram_version": 5,
"agent_sub_id": 0,
"agent": "10.1.2.1",
"packet_sequence_no": 54,
"sys_up_time": 3810403,
"samples_in_packet": 1,
"sample_type_tag": "0:2",
"sample_type": "counters_sample",
"sample_sequence_no": 5,
"source_id": "0:512",
"counter_block_tag": "0:2",
"if_index": 512,
"network_type": 6,
"if_speed": 10000000000,
"if_direction": 1,
"if_status": 3,
"if_in_octets": 0,
"if_in_ucast_pkts": 8,
"if_in_multicast_pkts": 121,
"if_in_broadcast_pkts": 0,
"if_in_discards": 0,
"if_in_errors": 0,
"if_in_unknown_protos": 0,
"if_out_octets": 130454,
"if_out_ucast_pkts": 1060,
"if_out_multicast_pkts": 0,
"if_out_broadcast_pkts": 95,
"if_out_discards": 0,
"if_out_errors": 0,
"if_promiscuous_mode": 0,
"dot3_stats_alignment_errors": 0,
"dot3_stats_fcserrors": 0,
"dot3_stats_single_collision_frames": 0,
"dot3_stats_multiple_collision_frames": 0,
"dot3_stats_sqetest_errors": 0,
"dot3_stats_deferred_transmissions": 0,
"dot3_stats_late_collisions": 0,
"dot3_stats_excessive_collisions": 0,
"dot3_stats_internal_mac_transmit_errors": 0,
"dot3_stats_carrier_sense_errors": 0,
"dot3_stats_frame_too_longs": 0,
"dot3_stats_internal_mac_receive_errors": 0,
"dot3_stats_symbol_errors": 0
}
Here is a quick benchmark on Macbook Pro Mid 2015:
Please report issues or enhancement requests to GitHub issues. For questions or feedbacks write to my twitter @codeout.
Or send a pull request to fix.
Copyright (c) 2017-2021 Shintaro Kojima. Code released under the MIT license.
Code includes a part of sflowtool which is distributed in the InMon sFlow License.