edeliver v1.2.6 Edeliver.Relup.Instructions.ResumeChannels

This upgrade instruction resumes the websocket processes

connected to phoenix channels when the upgrade is done to continue handling channel events. Use this instruction at the end of the upgrade modification if the

Edeliver.Relup.Instructions.SuspendChannels

is used at the beginning. Make sure that it is used before the

Edeliver.Relup.Instructions.ResumeRanchAcceptors

instruction to avoid that recently started websockets which were not suspendet are tried to be resumed.

Suspending and resuming websocket processes for phoenix channels requires a recent phoenix version which handles sys events for websockets. It also requires that the builtin phoenix pubsub backend Phoenix.PubSub.PG2 is used for the phoenix channels.

Summary

Types

insert_fun()

instruction()

instructions()

Functions

arguments(instructions, config)

Returns name of the application

bulk_resume(processes, timeout \\ 1000)

Resumes a list of processes

call_this(arguments \\ [])

Calls the run/1 function of this module

debug(message)

Logs a debug message using the Logger on the running node which is upgraded

dependencies()

This module depends on the Edeliver.Relup.Instructions.CheckRanchAcceptors and the Edeliver.Relup.Instructions.CheckRanchConnections module

ensure_dependencies_loaded_before_instruction_for_upgrade(instructions, call_this_instruction, dependencies \\ dependencies())

Ensures that all Edeliver.Relup.RunnableInstruction modules used / referenced by this instruction and returned by the dependencies/0 callback are loaded before this instruction is executed during the upgrade

ensure_dependencies_unloaded_after_instruction_for_downgrade(instructions, call_this_instruction, dependencies \\ dependencies())

Ensures that all Edeliver.Relup.RunnableInstruction modules used / referenced by this instruction and returned by the dependencies/0 callback are unloaded after this instruction is executed during the downgrade

error(message)

Logs an error using the Logger on the running node which is upgraded

format_in_upgrade_script(format, arguments)

Formats and prints the message on the node

info(message)

Logs an info message using the Logger on the running node which is upgraded

insert_where()

Callback implementation for Edeliver.Relup.RunnableInstruction.insert_where/0

log_in_upgrade_script(type, message)

Logs the message of the given type on the node

modify_relup(instructions, config)

Callback implementation for Edeliver.Relup.Instruction.modify_relup/2

run(otp_application_name)

Resumes all websocket channels

warn(message)

Logs a warning using the Logger on the running node which is upgraded

Macros

assume(assertion, error_message)

Assumes that the pattern matches or throws an error with the given error message

Types

insert_fun :: (%Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term} | instructions, new_instructions :: instruction | instructions -> updated_instructions :: %Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term} | instructions)

instruction :: :relup.instruction

instructions :: [instruction]

Functions

arguments(instructions, config)

Specs

arguments(%Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term}, %ReleaseManager.Config{dev: term, env: term, erl: term, name: term, package: term, relx_config: term, upgrade?: term, verbosity: term, version: term}) :: term

Returns name of the application.

This name is taken as argument for the run/1 function and is required to access the acceptor processes through the supervision tree

bulk_resume(processes, timeout \\ 1000)

Specs

bulk_resume(processes :: [pid], timeout :: pos_integer | :infinity) ::
  :ok |
  {:errors, count :: pos_integer, [{pid :: pid, reason :: term}]} |
  :not_supported

Resumes a list of processes.

Because resume a process might take a while depending on the length of the message queue or duration of current operation processed by the pid, suspending is done asynchronously for each process by spawing a new process which calls :sys.resume/2 and then waiting for all results before returning from this function. Be careful when using :infinity as timeout, because this function might hang for infinite time if one of the process does not handle sys events.

call_this(arguments \\ [])

Specs

call_this(arguments :: [term]) ::
  instruction |
  instructions

Calls the run/1 function of this module

from the relup file during hot code upgrade

debug(message)

Specs

debug(message :: String.t) :: no_return

Logs a debug message using the Logger on the running node which is upgraded.

In addition the same debug message is logged on the node which executes the upgrade and is displayed as output of the $APP/bin/$APP upgarde $RELEASE command.

dependencies()

Specs

dependencies :: [instruction_module :: atom]

dependencies :: [Edeliver.Relup.Instructions.CheckRanchAcceptors]

This module depends on the Edeliver.Relup.Instructions.CheckRanchAcceptors and the Edeliver.Relup.Instructions.CheckRanchConnections module

which must be loaded before this instruction for upgrades and unloaded after this instruction for downgrades.

ensure_dependencies_loaded_before_instruction_for_upgrade(instructions, call_this_instruction, dependencies \\ dependencies())

Specs

ensure_dependencies_loaded_before_instruction_for_upgrade(instructions :: %Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term}, runnable_instruction :: {:apply, {module :: atom, :run, arguments :: [term]}}, dependencies :: [instruction_module :: atom]) :: %Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term}

ensure_dependencies_unloaded_after_instruction_for_downgrade(instructions, call_this_instruction, dependencies \\ dependencies())

Specs

ensure_dependencies_unloaded_after_instruction_for_downgrade(instructions :: %Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term}, runnable_instruction :: {:apply, {module :: atom, :run, arguments :: [term]}}, dependencies :: [instruction_module :: atom]) :: %Edeliver.Relup.Instructions{changed_modules: term, down_instructions: term, down_version: term, up_instructions: term, up_version: term}

error(message)

Specs

error(message :: String.t) :: no_return

Logs an error using the Logger on the running node which is upgraded.

In addition the same error message is logged on the node which executes the upgrade and is displayed as output of the $APP/bin/$APP upgarde $RELEASE command.

format_in_upgrade_script(format, arguments)

Specs

format_in_upgrade_script(format :: char_list, arguments :: [term]) :: no_return

Formats and prints the message on the node

running the upgrade script which was started by the $APP/bin/$APP upgrade $RELEASE command.

info(message)

Specs

info(message :: String.t) :: no_return

Logs an info message using the Logger on the running node which is upgraded.

In addition the same info message is logged on the node which executes the upgrade and is displayed as output of the $APP/bin/$APP upgarde $RELEASE command.

insert_where()

Specs

insert_where :: insert_fun

Callback implementation for Edeliver.Relup.RunnableInstruction.insert_where/0.

log_in_upgrade_script(type, message)

Specs

log_in_upgrade_script(type :: :error | :warning | :info | :debug, message :: String.t) :: no_return

Logs the message of the given type on the node

which executes the upgrade and displays it as output of the $APP/bin/$APP upgrade $RELEASE command. The message is prefixed with a string derived from the message type.

modify_relup(instructions, config)

Callback implementation for Edeliver.Relup.Instruction.modify_relup/2.

run(otp_application_name)

Specs

run(otp_application_name :: atom) :: :ok

Resumes all websocket channels

to continue handling channel events after the upgrade. This is possible only in recent phoenix versions since handling sys events is required for resuming. If an older version is used, a warning is printed that suspending is not supported.

warn(message)

Specs

warn(message :: String.t) :: no_return

Logs a warning using the Logger on the running node which is upgraded.

In addition the same warning message is logged on the node which executes the upgrade and is displayed as output of the $APP/bin/$APP upgarde $RELEASE command.

Macros

assume(assertion, error_message)

Assumes that the pattern matches or throws an error with the given error message.

The error message is logged as error to the logfile using the Logger and displayed as error output by the $APP/bin/$APP upgrade $RELEASE task using the $APP/ebin/install_upgrade.escript script. If the pattern matches the variables from the matching are assigned.