Files
kumomta/crates/integration-tests/source.lua
T
Wez Furlong 7f4a929ca2 message transfer (xfer) support
This commit implements a kumomta-specific message transfer
protocol that is intended to be used to migrate messages
from one kumomta node to another.

The transfer is carried out using an HTTP POST request
to the destination node's http listener.

The request includes the full message metadata and body,
in a compressed form.

An xfer request can be made via `kcli xfer` (and thus also via an HTTP API
endpoint).  It works similarly to a rebind operation; you specify the
criteria to be used to match scheduled queues, along with the target
node for the xfer, and kumomta will find matching queues, drain out the
messages, make an adjustment to the metadata to capture current
scheduling information, and then place the messages into an xfer queue.

The xfer queue has hard-coded scheduling queue configuration of its own,
with the base retry interval set to 10 seconds, which should be suitably
aggressive for the intended use case.

You may apply shaping to affect the number of concurrent requests in a
similar way to how TSA shaping is configured.

On the receiving side, the incoming xfer sanity check to prohibit
trying to xfer to itself.

The spool id of the Message is not suitable to be reused verbatim on
another node (spool ids include the local mac address and creation
timestamp information, as well as a random component), so the receiving
side will derive an id that should be suitable for use on that node.
The originating node id and spool id will be preserved in metadata to
aid in tracing.

It is possible for an xfer request to target an existing xfer queue, so
that you can correct/update the target in various circumstances. In that
situation the messages will be "simply" moved from the source queue to
the destination queue.

It is possible to cancel an xfer request via `kcli xfer-cancel` (and
thus also via an HTTP API endpoint).  You specify the target queue,
which must be an xfer queue, and it will have its messages drained and
the metadata changes that were applied when the xfer was initiated will
be reversed, allowing the messages to then be reinserted into their
originating queue.

refs: https://github.com/KumoCorp/kumomta/issues/311
2025-10-07 10:58:07 +01:00

417 lines
12 KiB
Lua

local kumo = require 'kumo'
package.path = '../../assets/?.lua;' .. package.path
log_hooks = require 'policy-extras.log_hooks'
local TEST_DIR = os.getenv 'KUMOD_TEST_DIR'
local SINK_PORT = tonumber(os.getenv 'KUMOD_SMTP_SINK_PORT')
local WEBHOOK_PORT = os.getenv 'KUMOD_WEBHOOK_PORT'
local AMQPHOOK_URL = os.getenv 'KUMOD_AMQPHOOK_URL'
local AMQP_HOST_PORT = os.getenv 'KUMOD_AMQP_HOST_PORT'
local LISTENER_MAP = os.getenv 'KUMOD_LISTENER_DOMAIN_MAP'
local DEFERRED_SMTP_SERVER_MSG_INJECT =
os.getenv 'KUMOD_DEFERRED_SMTP_SERVER_MSG_INJECT'
local BATCH_HANDLING = (os.getenv 'KUMOD_BATCH_HANDLING') or 'BifurcateAlways'
local MAX_RECIPIENTS_PER_BATCH = (os.getenv 'KUMOD_MAX_RECIPIENTS_PER_BATCH')
or 100
local USE_SPLIT_TXN = os.getenv 'KUMOD_USE_SPLIT_TXN'
kumo.on('init', function()
kumo.configure_accounting_db_path(TEST_DIR .. '/accounting.db')
local relay_hosts = { '0.0.0.0/0' }
local RELAY_HOSTS = os.getenv 'KUMOD_RELAY_HOSTS'
if RELAY_HOSTS then
relay_hosts = kumo.json_parse(RELAY_HOSTS)
end
kumo.start_esmtp_listener {
listen = '127.0.0.1:0',
relay_hosts = relay_hosts,
deferred_queue = (DEFERRED_SMTP_SERVER_MSG_INJECT and true) or false,
peer = {
['127.0.0.1'] = {
allow_xclient = true,
},
},
via = {
['42.42.42.42'] = {
banner = 'what do you get when you multiply six by nine?',
},
},
batch_handling = BATCH_HANDLING,
}
kumo.start_http_listener {
listen = '127.0.0.1:0',
}
kumo.configure_local_logs {
log_dir = TEST_DIR .. '/logs',
max_segment_duration = '1s',
headers = { 'X-*', 'Y-*', 'Subject' },
meta = { 'xfer*' },
}
if WEBHOOK_PORT then
kumo.configure_log_hook {
name = 'webhook',
headers = { 'Subject', 'X-*' },
}
end
kumo.define_spool {
name = 'data',
path = TEST_DIR .. '/data-spool',
}
kumo.define_spool {
name = 'meta',
path = TEST_DIR .. '/meta-spool',
}
end)
if AMQPHOOK_URL then
log_hooks:new {
name = 'amqp',
constructor = function(domain, tenant, campaign)
local sender = {}
local client = kumo.amqp.build_client(AMQPHOOK_URL)
function sender:send(msg)
local result = client:publish_with_timeout({
routing_key = 'woot',
payload = msg:get_data(),
}, 20000)
if result.status == 'Ack' or result.status == 'NotRequested' then
return string.format('250 %s', kumo.json_encode(result))
end
-- result.status must be `Nack`; log the full result
kumo.reject(500, kumo.json_encode(result))
end
function sender:close()
client:close()
end
return sender
end,
}
elseif AMQP_HOST_PORT then
log_hooks:new {
name = 'amqp',
constructor = function(domain, tenant, campaign)
local sender = {}
local host, port = table.unpack(kumo.string.split(AMQP_HOST_PORT, ':'))
function sender:send(msg)
kumo.amqp.basic_publish {
routing_key = 'woot',
payload = msg:get_data(),
connection = {
host = host,
port = tonumber(port),
},
}
return '250 ok'
end
return sender
end,
}
end
if WEBHOOK_PORT then
local min_batch_size = tonumber(os.getenv 'KUMOD_WEBHOOK_MIN_BATCH_SIZE')
local max_batch_size = tonumber(os.getenv 'KUMOD_WEBHOOK_MAX_BATCH_SIZE')
local max_batch_latency = os.getenv 'KUMOD_WEBHOOK_MAX_BATCH_LATENCY'
if max_batch_size > 1 then
log_hooks:new {
name = 'webhookbatch',
batch_size = max_batch_size,
min_batch_size = min_batch_size,
max_batch_latency = max_batch_latency,
constructor = function(domain, tenant, campaign)
local sender = {}
local client = kumo.http.build_client {}
function sender:send_batch(messages)
local payload = {}
for _, msg in ipairs(messages) do
table.insert(payload, msg:get_meta 'log_record')
end
print(
string.format(
'batch size is %d *************** min=%d max=%d latency=%s',
#payload,
min_batch_size,
max_batch_size,
max_batch_latency
)
)
local response = client
:post(
string.format('http://127.0.0.1:%d/log-batch', WEBHOOK_PORT)
)
:header('Content-Type', 'application/json')
:body(kumo.serde.json_encode(payload))
:send()
local disposition = string.format(
'%d %s: %s',
response:status_code(),
response:status_reason(),
response:text()
)
if response:status_is_success() then
return disposition
end
kumo.reject(500, disposition)
end
return sender
end,
}
else
kumo.on('should_enqueue_log_record', function(msg)
local log_record = msg:get_meta 'log_record'
-- avoid an infinite loop caused by logging that we logged
if log_record.queue ~= 'webhook' then
msg:set_meta('queue', 'webhook')
return true
end
return false
end)
kumo.on('make.webhook', function(_domain, _tenant, _campaign)
local sender = {}
local client = kumo.http.build_client {}
function sender:send(message)
local response = client
:post(string.format('http://127.0.0.1:%d/log', WEBHOOK_PORT))
:header('Content-Type', 'application/json')
:body(message:get_data())
:send()
local disposition = string.format(
'%d %s: %s',
response:status_code(),
response:status_reason(),
response:text()
)
if response:status_is_success() then
return disposition
end
kumo.reject(500, disposition)
end
return sender
end)
end
end
kumo.on('get_listener_domain', function(domain, listener, conn_meta)
if LISTENER_MAP then
local map = kumo.json_parse(LISTENER_MAP)
local params = map[domain]
if params then
return kumo.make_listener_domain(params)
end
end
return kumo.make_listener_domain {
relay_to = true,
log_oob = true,
log_arf = true,
}
end)
if USE_SPLIT_TXN then
kumo.on('smtp_server_split_transaction', function(msg, conn)
-- This toy implementation batches by the first letter of
-- the local part and the same domain. It is NOT an example
-- of a valid production use case, it is merely exercising
-- the logic to prove that it is functioning
local batches = {}
local keyed = {}
for _, recip in ipairs(msg:recipient_list()) do
local key = recip.user:sub(1, 1) .. '@' .. recip.domain:lower()
if not keyed[key] then
keyed[key] = {}
end
table.insert(keyed[key], recip)
end
for _, batch in pairs(keyed) do
table.insert(batches, batch)
end
return batches
end)
end
kumo.on('smtp_server_message_received', function(msg)
local result = msg:import_scheduling_header 'X-Schedule'
kumo.log_info('schedule result', kumo.serde.json_encode(result))
-- This tenant header import is used by xfer.rs as a way to set
-- the tenant metadata for the incoming message
msg:import_x_headers { 'tenant' }
end)
kumo.on('get_queue_config', function(domain, tenant, campaign, routing_domain)
kumo.log_warn('get_queue_config', domain, tenant, campaign, routing_domain)
if domain == 'webhook' then
return kumo.make_queue_config {
protocol = {
custom_lua = {
constructor = 'make.webhook',
},
},
}
end
local protocol = {
-- Redirect traffic to the sink
smtp = {
mx_list = { 'localhost:' .. SINK_PORT },
},
}
if domain == 'two-hosts.example.com' then
protocol = {
-- Redirect traffic to the sink, but with two hosts in the connection plan
smtp = {
mx_list = {
{ name = 'localhost-1', addr = '127.0.0.1:' .. SINK_PORT },
{ name = 'localhost-2', addr = '127.0.0.1:' .. SINK_PORT },
},
},
}
end
if domain == 'nxdomain' then
-- this nxdomain domain is a special domain that is assumed not
-- to resolve. It is generated by the retry_schedule integration
-- test. for this domain, we don't want to short-circuit dns
-- and go to the sink, because we DO want the dns resolution
-- to successfully return nxdomain in order for the test to
-- exercise the appropriate logic.
protocol = nil
end
if domain == 'rebind.example.com' then
-- This domain is used by rebind_port.rs.
-- In this case, we don't want the default sink route that was
-- established above to take effect, we want to confirm that
-- the port number embedded in either the domain or routing_domain
-- are used, so we use the default smtp protocol settings here
-- and rely on assertions in rebind_port.rs to validate the
-- behavior.
protocol = nil
end
return kumo.make_queue_config {
protocol = protocol,
retry_interval = os.getenv 'KUMOD_RETRY_INTERVAL',
strategy = os.getenv 'KUMOD_QUEUE_STRATEGY',
egress_pool = os.getenv 'KUMOD_POOL_NAME',
}
end)
kumo.on('get_egress_pool', function(pool_name)
if pool_name == 'warming' then
-- coupled with source_selection_rate_pool.rs
return kumo.make_egress_pool {
name = pool_name,
entries = {
{ name = 'warming_a' },
{ name = 'warming_b' },
},
}
end
error('integration-tests/source.lua: unhandled pool ' .. pool_name)
end)
kumo.on('get_egress_source', function(source_name)
return kumo.make_egress_source {
name = source_name,
}
end)
kumo.on('get_egress_path_config', function(domain, source_name, _site_name)
-- Allow sending to a sink
local params = {
enable_tls = os.getenv 'KUMOD_ENABLE_TLS' or 'OpportunisticInsecure',
reconnect_strategy = os.getenv 'KUMOD_RECONNECT_STRATEGY'
or 'ConnectNextHost',
prohibited_hosts = {},
opportunistic_tls_reconnect_on_failed_handshake = (
(os.getenv 'KUMOD_OPPORTUNISTIC_TLS_RECONNECT') and true
) or false,
source_selection_rate = os.getenv 'KUMOD_SOURCE_SELECTION_RATE',
try_next_host_on_transport_error = (
(os.getenv 'KUMOD_TRY_NEXT_HOST_ON_TRANSPORT_ERROR') and true
) or false,
tls_prefer_openssl = ((os.getenv 'KUMOD_PREFER_OPENSSL') and true)
or false,
max_recipients_per_batch = tonumber(MAX_RECIPIENTS_PER_BATCH),
}
if os.getenv 'KUMOD_CLIENT_CERTIFICATE' then
params.tls_certificate = {
key_data = os.getenv 'KUMOD_CLIENT_CERTIFICATE',
}
end
if os.getenv 'KUMOD_CLIENT_PRIVATE_KEY' then
params.tls_private_key = {
key_data = os.getenv 'KUMOD_CLIENT_PRIVATE_KEY',
}
end
-- See if there is a source-specific rate exported to us via the environment.
-- We assign this using additional_source_selection_rates regardless of
-- whether we have a more generate rate specified above so that we can
-- excercise the additional_source_selection_rates collection logic in the core.
local source_rate_name = 'KUMOD_SOURCE_SELECTION_RATE_'
.. source_name:upper()
local source_rate = os.getenv(source_rate_name)
if source_rate then
params.additional_source_selection_rates =
{ [source_rate_name] = source_rate }
end
local username = os.getenv 'KUMOD_SMTP_AUTH_USERNAME'
local password = os.getenv 'KUMOD_SMTP_AUTH_PASSWORD'
if username and password then
params.smtp_auth_plain_username = username
params.smtp_auth_plain_password = {
key_data = password,
}
end
if domain == 'two-hosts.example.com' then
params.connection_limit = 1
end
if domain == 'webhookbatch.log_hook' then
-- If we allow more connections, then we can end up
-- with batches smaller than desired because we split
-- them among multiple connections
params.connection_limit = 1
end
kumo.log_warn('get_egress_path_config *******************', domain)
return kumo.make_egress_path(params)
end)
if os.getenv 'KUMOD_WANT_REBIND' then
kumo.on('rebind_message', function(message, data)
message:set_meta('queue', data.queue)
end)
end