storj/satellite/audit/reverifyworker.go
paul cannon c575a21509 satellite/audit: add audit.ReverifyWorker
Here we add a worker class comparable to audit.Worker, which will be
responsible for pulling items off of the reverification queue and
calling reverifier.ReverifyPiece on them.

Note that piecewise reverification audits (which this will control) are
not yet being done. That is, nothing is being added to the
reverification queue at this point.

Refs: https://github.com/storj/storj/issues/5251
Change-Id: I94e28830e27caa49f2c8bd4a2336533e187ab69c
2022-12-12 11:57:08 +00:00

115 lines
3.2 KiB
Go

// Copyright (C) 2022 Storj Labs, Inc.
// See LICENSE for copying information.
package audit
import (
"context"
"time"
"go.uber.org/zap"
"storj.io/common/sync2"
)
// ReverifyWorker processes reverifications (retrying piece audits against nodes that timed out
// during a Verification).
type ReverifyWorker struct {
log *zap.Logger
queue ReverifyQueue
reverifier *Reverifier
reporter Reporter
Loop *sync2.Cycle
concurrency int
retryInterval time.Duration
}
// NewReverifyWorker creates a new ReverifyWorker.
func NewReverifyWorker(log *zap.Logger, queue ReverifyQueue, reverifier *Reverifier, reporter Reporter, config Config) *ReverifyWorker {
return &ReverifyWorker{
log: log,
queue: queue,
reverifier: reverifier,
reporter: reporter,
Loop: sync2.NewCycle(config.QueueInterval),
concurrency: config.ReverifyWorkerConcurrency,
retryInterval: config.ReverificationRetryInterval,
}
}
// Run runs a ReverifyWorker.
func (worker *ReverifyWorker) Run(ctx context.Context) (err error) {
defer mon.Task()(&ctx)(&err)
return worker.Loop.Run(ctx, func(ctx context.Context) (err error) {
err = worker.process(ctx)
if err != nil {
worker.log.Error("failure processing reverify queue", zap.Error(Error.Wrap(err)))
}
return nil
})
}
func (worker *ReverifyWorker) process(ctx context.Context) (err error) {
defer mon.Task()(&ctx)(&err)
limiter := sync2.NewLimiter(worker.concurrency)
defer limiter.Wait()
for {
// We start the timeout clock _before_ pulling the next job from
// the queue. This gives us the best chance of having this worker
// terminate and get cleaned up before another reverification
// worker tries to take over the job.
//
// (If another worker does take over the job before this worker
// has been cleaned up, it is ok; the downside should only be
// duplication of work and monkit stats.)
ctx, cancel := context.WithTimeout(ctx, worker.retryInterval)
reverifyJob, err := worker.queue.GetNextJob(ctx, worker.retryInterval)
if err != nil {
cancel()
if ErrEmptyQueue.Has(err) {
return nil
}
return err
}
started := limiter.Go(ctx, func() {
defer cancel()
logger := worker.log.With(
zap.Stringer("Segment StreamID", reverifyJob.Locator.StreamID),
zap.Uint64("Segment Position", reverifyJob.Locator.Position.Encode()),
zap.Stringer("Node ID", reverifyJob.Locator.NodeID),
zap.Int("Piece Number", reverifyJob.Locator.PieceNum))
worker.work(ctx, logger, reverifyJob)
})
if !started {
cancel()
return ctx.Err()
}
}
}
func (worker *ReverifyWorker) work(ctx context.Context, logger *zap.Logger, job *ReverificationJob) {
defer mon.Task()(&ctx)(nil)
logger.Debug("beginning piecewise audit")
outcome, reputation := worker.reverifier.ReverifyPiece(ctx, logger, &job.Locator)
logger.Debug("piecewise audit complete", zap.Int("outcome", int(outcome)))
err := worker.reporter.RecordReverificationResult(ctx, job, outcome, reputation)
if err != nil {
logger.Error("finished with audit, but failed to remove entry from queue", zap.Error(err))
}
}
// Close halts the worker.
func (worker *ReverifyWorker) Close() error {
worker.Loop.Close()
return nil
}