1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
#[cfg(all(not(feature = "debugger"), not(feature = "submission")))]
use std::marker::PhantomData;
use log::{debug, error, trace};
use flowcore::errors::*;
#[cfg(feature = "metrics")]
use flowcore::model::metrics::Metrics;
use flowcore::model::submission::Submission;
#[cfg(feature = "debugger")]
use crate::debugger::Debugger;
#[cfg(feature = "debugger")]
use crate::debugger_handler::DebuggerHandler;
use crate::dispatcher::Dispatcher;
use crate::job::Job;
use crate::run_state::RunState;
#[cfg(feature = "submission")]
use crate::submission_handler::SubmissionHandler;
/// The `Coordinator` is responsible for coordinating the dispatching of jobs (consisting
/// of a set of Input values and an Implementation of a Function) for execution,
/// gathering the resulting Outputs and distributing output values to other connected function's
/// Inputs.
///
/// It accepts Flows to be executed in the form of a `Submission` struct that has the required
/// information to execute the flow.
pub struct Coordinator<'a> {
/// A `Server` to communicate with clients
#[cfg(feature = "submission")]
submission_handler: &'a mut dyn SubmissionHandler,
/// Dispatcher to dispatch jobs for execution
dispatcher: Dispatcher,
#[cfg(feature = "debugger")]
/// A `Debugger` to communicate with debug clients
debugger: Debugger<'a>,
#[cfg(all(not(feature = "debugger"), not(feature = "submission")))]
_data: PhantomData<&'a Dispatcher>
}
impl<'a> Coordinator<'a> {
/// Create a new `coordinator` with `num_threads` local executor threads
pub fn new(
dispatcher: Dispatcher,
#[cfg(feature = "submission")] submitter: &'a mut dyn SubmissionHandler,
#[cfg(feature = "debugger")] debug_server: &'a mut dyn DebuggerHandler
) -> Result<Self> {
Ok(Coordinator {
#[cfg(feature = "submission")]
submission_handler: submitter,
dispatcher,
#[cfg(feature = "debugger")]
debugger: Debugger::new(debug_server),
#[cfg(all(not(feature = "debugger"), not(feature = "submission")))]
_data: PhantomData
})
}
/// Enter a loop - waiting for a submission from the client, or disconnection of the client
#[cfg(feature = "submission")]
pub fn submission_loop(
&mut self,
loop_forever: bool,
) -> Result<()> {
while let Some(submission) = self.submission_handler.wait_for_submission()? {
let _ = self.execute_flow(submission);
if !loop_forever {
break;
}
}
self.submission_handler.coordinator_is_exiting(Ok(()))
}
//noinspection RsReassignImmutable
/// Execute a flow by looping while there are jobs to be processed in an inner loop.
/// There is an outer loop for the case when you are using the debugger, to allow entering
/// the debugger when the flow ends and at any point resetting all the state and starting
/// execution again from the initial state
#[allow(unused_variables, unused_assignments, unused_mut)]
pub fn execute_flow(&mut self,
submission: Submission,) -> Result<()> {
self.dispatcher.set_results_timeout(submission.job_timeout)?;
let mut state = RunState::new(submission);
#[cfg(feature = "metrics")]
let mut metrics = Metrics::new(state.num_functions());
#[cfg(feature = "debugger")]
if state.submission.debug_enabled {
self.debugger.start();
}
let mut restart = false;
let mut display_next_output = false;
// This outer loop is just a way of restarting execution from scratch if the debugger requests it
'flow_execution:
loop {
state.init()?;
#[cfg(feature = "metrics")]
metrics.reset();
// If debugging - then prior to starting execution - enter the debugger
#[cfg(feature = "debugger")]
if state.submission.debug_enabled {
(display_next_output, restart) = self.debugger.wait_for_command(&mut state)?;
}
#[cfg(feature = "submission")]
self.submission_handler.flow_execution_starting()?;
'jobs: loop {
trace!("{}", state);
#[cfg(feature = "debugger")]
if state.submission.debug_enabled && self.submission_handler.should_enter_debugger()? {
(display_next_output, restart) = self.debugger.wait_for_command(&mut state)?;
if restart {
break 'jobs;
}
}
(display_next_output, restart) = self.dispatch_jobs(
&mut state,
#[cfg(feature = "metrics")]
&mut metrics,
)?;
if restart {
break 'jobs;
}
if state.number_jobs_running() > 0 {
match self.dispatcher.get_next_result() {
Ok(result) => {
let job;
(display_next_output, restart, job) = state.retire_job(
#[cfg(feature = "metrics")] &mut metrics,
result,
#[cfg(feature = "debugger")] &mut self.debugger,
)?;
#[cfg(feature = "debugger")]
if display_next_output {
(display_next_output, restart) =
self.debugger.job_done(&mut state, &job)?;
if restart {
break 'jobs;
}
}
}
Err(err) => {
error!("\t{}", err.to_string());
#[cfg(feature = "debugger")]
if state.submission.debug_enabled {
(display_next_output, restart) = self.debugger.error(
&mut state, err.to_string())?;
if restart {
break 'jobs;
}
}
}
}
}
if state.number_jobs_running() == 0 && state.number_jobs_ready() == 0 {
// execution is done - but not returning here allows us to go into debugger
// at the end of execution, inspect state and possibly reset and rerun
break 'jobs;
}
} // 'jobs loop end
// flow execution has ended
#[allow(clippy::collapsible_if)]
#[cfg(feature = "debugger")]
if !restart {
{
// If debugging then enter the debugger for a final time before ending flow execution
if state.submission.debug_enabled {
(display_next_output, restart) = self.debugger.execution_ended(&mut state)?;
}
}
}
// if no debugger then end execution always
// if a debugger - then end execution if the debugger has not requested a restart
if !restart {
break 'flow_execution;
}
}
#[cfg(feature = "metrics")]
metrics.set_jobs_created(state.get_number_of_jobs_created());
#[cfg(all(feature = "submission", feature = "metrics"))]
self.submission_handler.flow_execution_ended(&state, metrics)?;
#[cfg(all(feature = "submission", not(feature = "metrics")))]
self.submitter.flow_execution_ended(&state)?;
Ok(()) // Normal flow completion exit
}
// Dispatch as many jobs as possible for parallel execution.
// Return if the debugger is requesting (display output, restart)
fn dispatch_jobs(
&mut self,
state: &mut RunState,
#[cfg(feature = "metrics")] metrics: &mut Metrics,
) -> Result<(bool, bool)> {
let mut display_output = false;
let mut restart = false;
while let Some(job) = state.get_job() {
match self.dispatch_a_job(
job.clone(),
state,
#[cfg(feature = "metrics")]
metrics,
) {
Ok((display, rest)) => {
display_output = display;
restart = rest;
}
Err(err) => {
error!("Error sending on 'job_tx': {}", err.to_string());
debug!("{}", state);
#[cfg(feature = "debugger")]
return self.debugger.job_error(state, &job); // TODO avoid above clone() ?
}
}
}
Ok((display_output, restart))
}
// Dispatch a job for execution
fn dispatch_a_job(
&mut self,
job: Job,
state: &mut RunState,
#[cfg(feature = "metrics")] metrics: &mut Metrics,
) -> Result<(bool, bool)> {
#[cfg(not(feature = "debugger"))]
let debug_options = (false, false);
#[cfg(feature = "metrics")]
metrics.track_max_jobs(state.number_jobs_running());
#[cfg(feature = "debugger")]
let debug_options = self
.debugger
.check_prior_to_job(state, &job)?;
self.dispatcher.send_job_for_execution(&job.payload)?;
state.start_job(job)?;
Ok(debug_options)
}
}