stompclt - Man Page

stompclt [Options]

stompclt is a versatile tool to interact with messaging brokers speaking STOMP and/or message queues (see Messaging::Message::Queue) on disk.

It receives messages (see Messaging::Message) from an incoming module, optionally massaging them (i.e. filtering and/or modifying), and sends them to an outgoing module. Depending on which modules are used, the tool can perform different operations.

Here are the supported incoming modules:

• broker: connect to a messaging broker using STOMP, subscribe to one or more destinations and receive the messages sent by the broker
• queue: read messages from a message queue on disk

Here are the supported outgoing modules:

• broker: connect to a messaging broker using STOMP and send the messages
• queue: store the messages in a message queue on disk

Here are some frequently used combinations:

• incoming broker + outgoing queue: drain some destinations, storing the messages on disk
• incoming queue + outgoing broker: (re-)send messages that have been previously stored on disk, optionally with modifications (such as altering the destination)
• incoming broker + outgoing broker: shovel messages from one broker to another

See the "Examples" sections for concrete examples.

--callback-code CODE

execute the given Perl code on each message, see the "Callback" section below for more information

--callback-data KEY=VALUE...

pass this data to the user supplied callback code, see the "Callback" section below for more information

--callback-path PATH

execute the Perl code in the given file on each message, see the "Callback" section below for more information

--config,  --conf,  --cfg PATH

use the given configuration file, see the "Configuration File" section below for more information

--config-general KEY=VALUE...

use the given Config::General options when creating the configuration parser

--count,  -c INTEGER

process at most the given number of messages; note: when using an incoming broker, to avoid consuming more messages, it is recommended to enable the --reliable option

--daemon | --no-daemon

detach stompclt so that it becomes a daemon running in the background; debug, warning and error messages will get sent to syslog; this option can be negated

--debug,  -d

show debugging information

--duration SECONDS

process messages during at most the given number of seconds and then stop; can be fractional

--heart-beat | --no-heart-beat

enable STOMP 1.1 heart-beats between stompclt and the broker(s); this option can be negated

--help,  -h,  -?

show some help

--incoming-broker-auth STRING

use this authentication string (see Authen::Credential) to authenticate to the incoming broker; this option can be given multiple times

--incoming-broker-connect KEY=VALUE...

use these options in the STOMP CONNECT frame sent to the incoming broker

--incoming-broker-sockopts KEY=VALUE...

use these socket options when connecting to the incoming broker

--incoming-broker-stomp-debug STRING

set the STOMP debug flags (see Net::STOMP::Client) when interacting with the incoming broker

--incoming-broker-type STRING

set the incoming broker type; this can be useful when using STOMP features which are broker specific

--incoming-broker-uri URI

use this connection URI (see Net::STOMP::Client) to connect to the incoming broker

--incoming-queue KEY=VALUE...

read incoming messages from the given message queue (see Messaging::Message::Queue)

--lazy | --no-lazy

initialize the outgoing module only after having received the first message; this option can be negated

--list,  -l

show all supported options

--loop | --no-loop

when using an incoming message queue, loop over it; this option can be negated

--manual,  -m

show this manual

--outgoing-broker-auth STRING

use this authentication string (see Authen::Credential) to authenticate to the outgoing broker; this option can be given multiple times

--outgoing-broker-connect KEY=VALUE...

use these options in the STOMP CONNECT frame sent to the outgoing broker

--outgoing-broker-sockopts KEY=VALUE...

use these socket options when connecting to the outgoing broker

--outgoing-broker-stomp-debug STRING

set the STOMP debug flags (see Net::STOMP::Client) when interacting with the outgoing broker

--outgoing-broker-type STRING

set the outgoing broker type; this can be useful when using STOMP features which are broker specific

--outgoing-broker-uri URI

use this connection URI (see Net::STOMP::Client) to connect to the outgoing broker

--outgoing-queue KEY=VALUE...

store outgoing messages into the given message queue (see Messaging::Message::Queue)

--pidfile PATH

use this pid file

--prefetch INTEGER

set the prefetch value (i.e. the maximum number of messages to received without acknowledging them) on the incoming broker

--quit

tell another instance of stompclt (identified by its pid file, as specified by the --pidfile option) to quit

--reliable | --no-reliable

use STOMP features for more reliable messaging (i.e. client side acknowledgments and receipts) at the cost of less performance; this option can be negated

--remove | --no-remove

when using an incoming message queue, remove the processed messages; this option can be negated

--statistics,  -s | --no-statistics

report statistics at the end of the execution; this option can be negated

--status

get the status of another instance of stompclt (identified by its pid file, as specified by the --pidfile option); the exit code will be zero if the instance is alive and non-zero otherwise

--subscribe KEY=VALUE...

use these options in the STOMP SUBSCRIBE frame used with the incoming broker; this option can be given multiple times

--timeout-broker SECONDS

use this timeout when interacting with the broker (e.g. getting receipts back); can be fractional

--timeout-client SECONDS

use this timeout for the client heart-beat; can be fractional (default: 40)

--timeout-connect SECONDS

use this timeout when connecting to the broker; can be fractional

--timeout-disconnect SECONDS

use this timeout when disconnecting from the broker; can be fractional (default: 60)

--timeout-flush SECONDS

use this timeout when attempting to send the last bytes to the broker just before disconnecting; can be fractional (default: 60)

--timeout-inactivity SECONDS

use this timeout in the incoming module to stop stompclt when no new messages have been received (aka drain mode); can be fractional

--timeout-linger SECONDS

when stopping stompclt, use this timeout to finish interacting with the broker; can be fractional

--timeout-server SECONDS

use this timeout for the server heart-beat; can be fractional (default: 10)

--timeout-status SECONDS

use this timeout when checking the status with --status; can be fractional

--unsubscribe KEY=VALUE...

use these options in the STOMP UNSUBSCRIBE frame used with the incoming broker; this option can be given multiple times and should match the --subscribe options

--version

display version information

--window INTEGER

keep at most the given number of not-yet-acknowledged messages in memory

To list all the available options in a compact form, type:

  $ stompclt -l

stompclt can read its options from a configuration file. For this, the Config::General module is used and the option names are the same as on the command line. For instance:

  daemon = true
  pidfile = /var/run/stompclt.pid
  incoming-queue = path=/var/spool/stompclt
  outgoing-broker-uri = stomp://broker.acme.com:6163
  outgoing-broker-auth = "plain name=guest pass=guest"

Alternatively, options can be nested:

  <outgoing-broker>
      uri = stomp://broker.acme.com:6163
      auth = "plain name=guest pass=guest"
  </outgoing-broker>

Or even:

  <outgoing>
      <broker>
          uri = stomp://broker.acme.com:6163
          <auth>
              scheme = plain
              name = guest
              pass = guest
          </auth>
      </broker>
  </outgoing>

The options specified on the command line have precedence over the ones found in the configuration file.

stompclt can be given Perl code to execute on all processed messages. This can be used for different purposes:

• massaging: the code can change any part of the message, including setting or removing header fields
• filtering: the code can decide if the message must be given to the outgoing module or not
• displaying: the code can print any part of the message
• copying: the code can store a copy of the message into files or message queues

To use callbacks, the --callback-path or --callback-code option must be used. The Perl code must provide functions with the following signature:

start(DATA)

(optional) this will be called when the program starts, with the supplied data (see the --callback-data option) as a hash reference

check(MESSAGE)

(mandatory) this will be called when the program has one message to process; it will be given the message (see Messaging::Message) and must return either a message (it could be the same one or a new one) or a string describing why the message has been dropped

idle()

(optional) this will be called when the program has no message to process

stop()

(optional) this will be called when the program stops

The code can be put in a file, on the command line or in the stompclt configuration file, using the "here document" syntax.

Here is an example (to be put in the stompclt configuration file) that prints on stdout a JSON array of messages:

  callback-code = <<EOF
      my($count);
      sub start ($) {
          $count = 0;
      }
      sub check ($) {
          my($msg) = @_;
          print($count++ ? "," : "[");
          print($msg->serialize(), "\n");
          return($msg);
      }
      sub stop () {
          print($count ? "]\n" : "[]\n");
      }
  EOF

For simple callback code that only needs the check subroutine, it is enough to supply the "inside code". If the subroutine definition is missing, the supplied code will be wrapped with:

  sub check ($) {
      my($msg) = @_;
      local *hdr = $msg->header();
      local *bdy = $msg->body_ref();
      ... your code goes here ...
      return($msg);
  }

This allows for instance to remove the message-id header with something like:

  $ stompclt ... --callback-code 'delete($hdr{"message-id"})'

Or to filter on message bodies with:

  $ stompclt ... --callback-code 'return("skip") unless $bdy =~ /error/'

In the case of an incoming broker, stompclt deals with the subscriptions defined by the --subscribe option.

Regardless of the --reliable option, subscriptions are always made using receipts. Also, if missing, an id header is always added.

Here is for instance how to create a named durable topic subscription using Apollo:

  $ stompclt ... --subscribe 'destination=/topic/foo persistent=true id=mysub'

By default, when it finishes, stompclt does not unsubscribe. It simply disconnects from the broker and the latter will perform the necessary cleanup when terminating the STOMP connection.

If the --unsubscribe option is given, even if it is empty, stompclt will explicitly unsubscribe before disconnecting, also using receipts.

Here is for instance how to destroy, when stompclt ends, the durable topic subscription created above:

  $ stompclt ... --unsubscribe 'persistent=true'

There is no need to give the subscription id in the --unsubscribe option because, by default, it comes from the matching --subscribe option.

stompclt has experimental UDP support (outgoing only). This has been tested with Apollo.

To use it, simply specify an outgoing URI that uses UDP such as:

  $ stompclt ... --outgoing-broker-uri udp://broker.acme.com:6163

Features such as authentication, heart beating, reliability and socket options are not supported over UDP.

Authen::Credential, Config::General, Messaging::Message, Messaging::Message::Queue, Net::STOMP::Client.

Lionel Cons <http://cern.ch/lionel.cons>

Copyright (C) CERN 2012-2021