The copy output plugin copies events to multiple outputs.
It is included in Fluentd's core.
<match pattern>
@type copy
<store>
@type file
path /var/log/fluent/myapp1
...
</store>
<store>
...
</store>
<store>
...
</store>
</match>
Please see the Configuration File article for the basic structure and syntax of the configuration file.
Here is an example set up to send events to both a local file under /var/log/fluent/myapp and the collection fluentd.test to an Elasticsearch instance (See out_file and out_elasticsearch):
<match myevent.file_and_elasticsearch>
@type copy
<store>
@type file
path /var/log/fluent/myapp
compress gzip
<format>
localtime false
</format>
<buffer time>
timekey_wait 10m
timekey 86400
timekey_use_utc true
path /var/log/fluent/myapp
</buffer>
<inject>
time_format %Y%m%dT%H%M%S%z
localtime false
</inject>
</store>
<store>
@type elasticsearch
host fluentd
port 9200
index_name fluentd
type_name fluentd
</store>
</match>
The value must be copy.
| type | default | available | version |
|---|---|---|---|
| enum | no_copy | no_copy, shallow, deep, marshal | 1.8.1 |
Chooses how to pass the events to <store> plugins.
Supported modes:
-
no_copy(default)Share events between
storeplugins. -
shallowPass shallow copied events to each
storeplugin. This mode uses Ruby'sdupmethod. This mode is useful when you do not modify the nested fields afterout_copy, e.g. remove top-level fields. -
deepPass deep copied events to each
storeplugin. This mode usesmsgpack-rubyinternally. This mode is useful when you modify the nested field afterout_copy, e.g. Kubernetes related fields. -
marshalPass deep copied events to each
storeplugin. This mode uses Ruby'smarshalinternally. This mode is useful whenmsgpack-rubycannot process your events. This mode is very slow.
| type | default | version |
|---|---|---|
| bool | false | 0.14.0 |
This parameter is deprecated since v1.8.1. Use copy_mode instead.
out_copy shares a record between store plugins by default.
If true, out_copy passes dupped record to each store plugin. This behavior is similar to copy_mode shallow.
Specifies the storage destinations. The format is the same as the <match> directive.
This section is required at least once.
If one store raises an error, it affects other <store>. For example:
<match app.**>
@type copy
<store>
@type plugin1
</store>
<store>
@type plugin2
</store>
</match>
If plugin1's emit/format raises an error, plugin2 is not executed. If you want to ignore an error from a less important <store>, you can specify ignore_error in <store>:
<match app.**>
@type copy
<store ignore_error>
@type plugin1
</store>
<store>
@type plugin2
</store>
</match>
Since Fluentd v1.12.2, you can use ignore_if_prev_success to define fallback outputs. For example:
<match app.**>
@type copy
<store ignore_error>
@type plugin1
name c0
</store>
<store ignore_if_prev_success ignore_error>
@type plugin2
name c1
</store>
</match>
Fluentd will make use of plugin2 only if the preceding destinations (plugin1 in this case) fail.
If this article is incorrect or outdated, or omits critical information, please let us know. Fluentd is an open-source project under Cloud Native Computing Foundation (CNCF). All components are available under the Apache 2 License.
