Example #1
0
func (c *Crawler) Start(states file.States) error {

	logp.Info("Loading Prospectors: %v", len(c.prospectorConfigs))

	// Prospect the globs/paths given on the command line and launch harvesters
	for _, prospectorConfig := range c.prospectorConfigs {

		prospector, err := prospector.NewProspector(prospectorConfig, states, c.spooler.Channel)
		if err != nil {
			return fmt.Errorf("Error in initing prospector: %s", err)
		}
		c.prospectors = append(c.prospectors, prospector)
	}

	logp.Info("Loading Prospectors completed. Number of prospectors: %v", len(c.prospectors))

	for i, p := range c.prospectors {
		c.wg.Add(1)

		go func(id int, prospector *prospector.Prospector) {
			defer func() {
				c.wg.Done()
				logp.Debug("crawler", "Prospector %v stopped", id)
			}()
			logp.Debug("crawler", "Starting prospector %v", id)
			prospector.Run()
		}(i, p)
	}

	logp.Info("All prospectors are initialised and running with %d states to persist", states.Count())

	return nil
}
Example #2
0
func NewProspector(cfg *common.Config, states file.States, spoolerChan chan *input.FileEvent) (*Prospector, error) {
	prospector := &Prospector{
		cfg:           cfg,
		config:        defaultConfig,
		spoolerChan:   spoolerChan,
		harvesterChan: make(chan *input.FileEvent),
		done:          make(chan struct{}),
		states:        states.Copy(),
		wg:            sync.WaitGroup{},
	}

	if err := cfg.Unpack(&prospector.config); err != nil {
		return nil, err
	}
	if err := prospector.config.Validate(); err != nil {
		return nil, err
	}

	err := prospector.Init()
	if err != nil {
		return nil, err
	}

	logp.Debug("prospector", "File Configs: %v", prospector.config.Paths)

	return prospector, nil
}
Example #3
0
func NewProspector(cfg *common.Config, states file.States, outlet Outlet) (*Prospector, error) {
	prospector := &Prospector{
		cfg:           cfg,
		config:        defaultConfig,
		outlet:        outlet,
		harvesterChan: make(chan *input.Event),
		done:          make(chan struct{}),
		wg:            sync.WaitGroup{},
		states:        &file.States{},
		channelWg:     sync.WaitGroup{},
	}

	if err := cfg.Unpack(&prospector.config); err != nil {
		return nil, err
	}
	if err := prospector.config.Validate(); err != nil {
		return nil, err
	}

	err := prospector.Init(states.GetStates())
	if err != nil {
		return nil, err
	}

	logp.Debug("prospector", "File Configs: %v", prospector.config.Paths)

	return prospector, nil
}
Example #4
0
// Init sets up the prospector
// It goes through all states coming from the registry. Only the states which match the glob patterns of
// the prospector will be loaded and updated. All other states will not be touched.
func (p *ProspectorLog) Init(states file.States) error {
	logp.Debug("prospector", "exclude_files: %s", p.config.ExcludeFiles)

	for _, state := range states.GetStates() {
		// Check if state source belongs to this prospector. If yes, update the state.
		if p.matchesFile(state.Source) {
			state.TTL = -1

			// Update prospector states and send new states to registry
			err := p.Prospector.updateState(input.NewEvent(state))
			if err != nil {
				logp.Err("Problem putting initial state: %+v", err)
				return err
			}
		}
	}

	logp.Info("Prospector with previous states loaded: %v", p.Prospector.states.Count())
	return nil
}