Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Feature: full backups #2710

Merged
merged 30 commits into from Nov 8, 2018
Merged

Feature: full backups #2710

merged 30 commits into from Nov 8, 2018

Conversation

srfrog
Copy link
Contributor

@srfrog srfrog commented Oct 30, 2018

This PR adds full backups to Dgraph.

Backups are started from any alpha using HTTP to /admin/backup via POST. A destination variable must be sent with the request to indicate where to save the backup.

curl  -XPOST -F 'destination=/some/path' localhost:8080/admin/backup

If the destination path must exist, except for local which will attempt to create it. Currently only local or NFS paths are added, but we will be adding AWS, GCP, Azure, and HTTP next.


This change is Reviewable

return
}
w.Header().Set("Content-Type", "application/json")
w.Write([]byte(`{"code": "Success", "message": "Backup completed."}`))

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Error return value of w.Write is not checked

worker/backup.go Outdated
}
readTs := ts.ReadOnly
glog.Infof("Got readonly ts from Zero: %d\n", readTs)
posting.Oracle().WaitForTs(ctx, readTs)

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Error return value of (*github.com/dgraph-io/dgraph/posting.oracle).WaitForTs is not checked

@@ -24,21 +24,27 @@ import (
"github.com/dgraph-io/badger"
"github.com/dgraph-io/dgraph/protos/pb"
"github.com/dgraph-io/dgraph/x"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

File is not goimports-ed

Copy link
Contributor

@manishrjain manishrjain left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewable status: 0 of 20 files reviewed, 19 unresolved discussions (waiting on @golangcibot, @srfrog, and @manishrjain)


ee/backup/backup.go, line 18 at r2 (raw file):

// Worker has all the information needed to perform a backup.
type Worker struct {

type Backup struct?


ee/backup/backup.go, line 21 at r2 (raw file):

	ReadTs    uint64     // Timestamp to read at.
	GroupId   uint32     // The group ID of this node.
	SeqTs     string     // Sequence data to label backup at the target.

UnixTs string // UTC, as we do in export.


ee/backup/backup.go, line 40 at r2 (raw file):

	sl.ItemToKVFunc = func(key []byte, itr *badger.Iterator) (*pb.KV, error) {
		item := itr.Item()
		val, err := item.ValueCopy(nil)

Look at predicate.go. We need to convert to posting list, and then marshal to KV.

if schema, then do value copy. Again, same logic as predicate.go.


ee/backup/handler.go, line 22 at r2 (raw file):

type handler interface {
	Copy(string, string) error
	Session(string, string) error

Send(kvs) error, so it would work with orchestrate.
Flush error


ee/backup/handler.go, line 32 at r2 (raw file):

//     s3 - Amazon S3
//     as - Azure Storage
var handlers struct {

Do we really need to store this at Dgraph? The user request can pass in the destination.


ee/backup/writer.go, line 65 at r2 (raw file):

	// tmp file is our main working file.
	// we will prepare this file and then copy to dst when done.
	w.tmp, err = ioutil.TempFile("", dgraphBackupTempPrefix)

We don't need to first write to a tmp file, then upload to destination. We can stream directly to the destination.


ee/backup/writer.go, line 86 at r2 (raw file):

	var err error
	for _, kv := range kvs.Kv {
		_, err = pbutil.WriteDelimited(w.tmp, kv)

Let's skip the lib. Just use fixed length header for the length of marshalled proto, then write the proto.


protos/pb.proto, line 444 at r2 (raw file):

message BackupRequest {
	uint64 start_ts = 1;

read_ts


protos/pb.proto, line 449 at r2 (raw file):

}

message BackupResponse {

I've never been happy with the ExportPayload, that this one is based on. Technically, all we need to pass back is an error.

message Status {
string err = 1; // If empty, then OK. If not, then it has the relevant error.
int exit_code = 2; // If 0, then OK. If 1, then error.
}

api.Payload, which can be used for this purpose as well. But, I think a special Status one is helpful.


worker/backup.go, line 78 at r2 (raw file):

// TODO: add stop to all goroutines to cancel on failure.
func backupDispatch(ctx context.Context, readTs uint64, target string, gids []uint32,

Move some args down, so the next line doesn't need to start with end-bracket?


worker/backup.go, line 79 at r2 (raw file):

// TODO: add stop to all goroutines to cancel on failure.
func backupDispatch(ctx context.Context, readTs uint64, target string, gids []uint32,
) chan *pb.BackupResponse {

This should just return error.


worker/backup.go, line 80 at r2 (raw file):

func backupDispatch(ctx context.Context, readTs uint64, target string, gids []uint32,
) chan *pb.BackupResponse {
	out := make(chan *pb.BackupResponse)

Out should not be returned. Rename to statusCh.

statusCh := make(chan *pb.Status, )

The function would just return error or status. That keeps all the goroutines, channels, etc. local to the function. We'd want a select case checking for context.done and the read from statusCh.


worker/backup.go, line 132 at r3 (raw file):

	// This will dispatch the request to all groups and wait for their response.
	// If we receive any failures, we cancel the process.
	for resp := range backupDispatch(ctx, readTs, target, gids) {

backupDispatch should return an error.


ee/backup/handler_file.go, line 37 at r3 (raw file):

// Copy is called when we are ready to transmit a file to the target.
// Returns error on failure, nil on success.
func (h *fileHandler) Copy(in, out string) error {

Send(kvs *pb.KVS) error

So, this can plug in to orchestrate. Also, have a Flush() method, which can be called at the end.


worker/stream/stream.go, line 33 at r3 (raw file):

)

const pageSize = 1 << 20 * 4 // 4MB

4 << 20


ee/backup/writer_local.go, line 65 at r2 (raw file):

	defer dst.Close()

	if _, err = io.Copy(dst, src); err != nil {

Why not just write directly to the destination?

Copy link
Contributor Author

@srfrog srfrog left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewable status: 0 of 20 files reviewed, 19 unresolved discussions (waiting on @golangcibot and @manishrjain)


dgraph/cmd/alpha/admin.go, line 87 at r1 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

Error return value of w.Write is not checked

We'll come back to this later


ee/backup/backup.go, line 18 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

type Backup struct?

backup.Backup{} looks weird. this is backup.Worker{}. I have changed it to backup.Request{}, maybe this looks better?


ee/backup/backup.go, line 21 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

UnixTs string // UTC, as we do in export.

Done.


ee/backup/backup.go, line 40 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Look at predicate.go. We need to convert to posting list, and then marshal to KV.

if schema, then do value copy. Again, same logic as predicate.go.

Done.


ee/backup/handler.go, line 22 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Send(kvs) error, so it would work with orchestrate.
Flush error

backup.writer implements kvStream, and imports the handler. i'll look to simply it after it's working.


ee/backup/handler.go, line 32 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Do we really need to store this at Dgraph? The user request can pass in the destination.

The user sends it in the HTTP request as destination.
e.g.,

curl -X POST -F 'destination=s3://dgraph.s3.amazonaws.com/backups/201810/' http://localhost:8080/admin/backup
# Dgraph will grab the AWS auth info from the env.

ee/backup/writer.go, line 65 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

We don't need to first write to a tmp file, then upload to destination. We can stream directly to the destination.

Done.


ee/backup/writer.go, line 86 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Let's skip the lib. Just use fixed length header for the length of marshalled proto, then write the proto.

Done.


protos/pb.proto, line 444 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

read_ts

Done.


protos/pb.proto, line 449 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

I've never been happy with the ExportPayload, that this one is based on. Technically, all we need to pass back is an error.

message Status {
string err = 1; // If empty, then OK. If not, then it has the relevant error.
int exit_code = 2; // If 0, then OK. If 1, then error.
}

api.Payload, which can be used for this purpose as well. But, I think a special Status one is helpful.

Done.


worker/backup.go, line 124 at r1 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

Error return value of (*github.com/dgraph-io/dgraph/posting.oracle).WaitForTs is not checked

Done.


worker/backup.go, line 78 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Move some args down, so the next line doesn't need to start with end-bracket?

Done.


worker/backup.go, line 79 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

This should just return error.

Done.


worker/backup.go, line 80 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Out should not be returned. Rename to statusCh.

statusCh := make(chan *pb.Status, )

The function would just return error or status. That keeps all the goroutines, channels, etc. local to the function. We'd want a select case checking for context.done and the read from statusCh.

Done.


worker/backup.go, line 132 at r3 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

backupDispatch should return an error.

Done.


ee/backup/handler_file.go, line 37 at r3 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Send(kvs *pb.KVS) error

So, this can plug in to orchestrate. Also, have a Flush() method, which can be called at the end.

i need to call this from writer, but i'll refactor if i can. i have Close() that does the final Flush()


worker/stream/stream.go, line 27 at r1 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

File is not goimports-ed

Done.


worker/stream/stream.go, line 33 at r3 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

4 << 20

Done.


ee/backup/writer_local.go, line 65 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Why not just write directly to the destination?

Done.

removed binary encoding package, using encode/binary with size delimiter.
fixed race and added posting list values to backup.
fixed issue with file handler that was breaking badger.
refactored backup process to be simpler.
added generic Status proto for any service response.
added minio and used it for S3 backup uploads.
"github.com/golang/glog"
)

const dgraphBackupTempPrefix = "dgraph-backup-*"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

dgraphBackupTempPrefix is unused

@@ -24,21 +24,24 @@ import (
"github.com/dgraph-io/badger"
"github.com/dgraph-io/dgraph/protos/pb"
"github.com/dgraph-io/dgraph/x"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

File is not goimports-ed

Copy link
Contributor Author

@srfrog srfrog left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewable status: 0 of 23 files reviewed, 21 unresolved discussions (waiting on @golangcibot and @manishrjain)


ee/backup/writer.go, line 17 at r4 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

dgraphBackupTempPrefix is unused

Done.


stream/stream.go, line 27 at r4 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

File is not goimports-ed

Done.

"github.com/dgraph-io/dgraph/protos/pb"
"github.com/dgraph-io/dgraph/stream"
"github.com/dgraph-io/dgraph/x"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

File is not goimports-ed

"path/filepath"

"github.com/dgraph-io/dgraph/x"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

File is not goimports-ed

"strings"

"github.com/dgraph-io/dgraph/x"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

File is not goimports-ed

Copy link
Contributor Author

@srfrog srfrog left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewable status: 0 of 23 files reviewed, 24 unresolved discussions (waiting on @golangcibot and @manishrjain)


dgraph/cmd/alpha/admin.go, line 87 at r1 (raw file):

Previously, srfrog (Gus) wrote…

We'll come back to this later

Done.


ee/backup/backup.go, line 18 at r2 (raw file):

Previously, srfrog (Gus) wrote…

backup.Backup{} looks weird. this is backup.Worker{}. I have changed it to backup.Request{}, maybe this looks better?

Done.


ee/backup/backup.go, line 16 at r5 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

File is not goimports-ed

Done.


ee/backup/handler.go, line 22 at r2 (raw file):

Previously, srfrog (Gus) wrote…

backup.writer implements kvStream, and imports the handler. i'll look to simply it after it's working.

Done.


ee/backup/handler.go, line 32 at r2 (raw file):

Previously, srfrog (Gus) wrote…

The user sends it in the HTTP request as destination.
e.g.,

curl -X POST -F 'destination=s3://dgraph.s3.amazonaws.com/backups/201810/' http://localhost:8080/admin/backup
# Dgraph will grab the AWS auth info from the env.

Done.


ee/backup/handler_s3.go, line 16 at r5 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

File is not goimports-ed

Done.


ee/backup/handler_file.go, line 37 at r3 (raw file):

Previously, srfrog (Gus) wrote…

i need to call this from writer, but i'll refactor if i can. i have Close() that does the final Flush()

Done.


ee/backup/handler_file.go, line 13 at r5 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

File is not goimports-ed

Done.

Copy link
Contributor

@manishrjain manishrjain left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The PR is a lot closer to submission. I'd like to change a couple of things here and there, see what code we can remove to slim it down a bit more. Let's chat after lunch.

Reviewed 3 of 19 files at r1, 3 of 13 files at r3, 12 of 17 files at r4, 5 of 5 files at r5.
Reviewable status: all files reviewed, 20 unresolved discussions (waiting on @golangcibot, @manishrjain, and @srfrog)


ee/backup/backup.go, line 63 at r5 (raw file):

	}

	glog.V(3).Infof("Backup started ...")

This can be V(2).


ee/backup/backup.go, line 67 at r5 (raw file):

		return err
	}
	glog.V(3).Infof("Backup finishing ...")

Can be V(2).


worker/backup.go, line 40 at r5 (raw file):

	// sanity, make sure this is our group.
	if groups().groupId() != req.GroupId {
		err := x.Errorf("Backup request group mismatch. Mine: %d. Requested: %d\n",

return x.Errorf


worker/backup.go, line 50 at r5 (raw file):

	// create backup request and process it.
	br := &backup.Request{
		ReadTs:    req.ReadTs,

Instead of copying, why not pass pb.BackupRequest to backup package.


worker/backup.go, line 77 at r5 (raw file):

func backupDispatch(ctx context.Context, in *pb.BackupRequest, gids []uint32) chan error {
	statusCh := make(chan error)
	go func() {

There's something not right about a function spawning a goroutine and just leaving them there. I find it easier to follow and a good practice when the func which spawns a goroutine also ensures that the goroutine reaches its conclusion.

This func backupDispatch, probably an apt name for what it's doing right now, should really do the dispatching, but also run through the results, returning a single error (not a chan).


worker/backup.go, line 132 at r5 (raw file):

	// If we receive any failures, we cancel the process.
	req := &pb.BackupRequest{ReadTs: readTs, Target: target}
	for err := range backupDispatch(ctx, req, gids) {

There';s


x/file.go, line 44 at r5 (raw file):

// Writeq is a quiet write. Writes b to w and eats up the return.
func Writeq(w io.Writer, b []byte) {
	_, _ = w.Write(b)

Can probably check? x.Check2(w.Write(b))


ee/backup/handler_file.go, line 35 at r5 (raw file):

		return err
	}
	glog.V(3).Infof("using file path: %q", path)

Can be V(2).


ee/backup/handler_file.go, line 47 at r5 (raw file):

		}
	}()
	if err := h.fp.Sync(); err != nil {

If Sync fails, then just do fp.Close().

But, do return the error of Close, so remove the defer, and bring it inline after Sync.

glog.V(3).Infof("sent %d bytes, actual %d bytes, time elapsed %s", n, h.n, time.Since(start))
break
}
pr.CloseWithError(nil) // EOF

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Error return value of pr.CloseWithError is not checked


const (
s3defaultEndpoint = "s3.amazonaws.com"
s3accelerateHost = "s3-accelerate"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s3accelerateHost is unused

"github.com/golang/glog"
)

const dgraphBackupSuffix = ".dgraph-backup"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

dgraphBackupSuffix is unused

renamed s3handler.send to s3handler.upload and removed all buffering.
s3handler tests that bucket exists before working, we cant assume a region.
s3handler.Close blocks until the upload is complete.

const (
s3DefaultEndpoint = "s3.amazonaws.com"
s3AccelerateHost = "s3-accelerate"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

s3AccelerateHost is unused

const (
s3DefaultEndpoint = "s3.amazonaws.com"
s3AccelerateHost = "s3-accelerate"
s3MinioChunkSize = 64 << 20 // 64MiB, minimum upload size for single file.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

const s3MinioChunkSize is unused


// progress allows us to monitor the progress of an upload.
// TODO: I used this during testing, maybe keep it turned on for -v 5 ?
type progress struct{ n uint64 }

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

type progress is unused

Copy link
Contributor Author

@srfrog srfrog left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewable status: 12 of 20 files reviewed, 26 unresolved discussions (waiting on @manishrjain and @golangcibot)


ee/backup/backup.go, line 63 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

This can be V(2).

Done.


ee/backup/backup.go, line 67 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Can be V(2).

Done.


ee/backup/handler_s3.go, line 26 at r6 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

s3accelerateHost is unused

Done.


ee/backup/handler_s3.go, line 107 at r6 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

Error return value of pr.CloseWithError is not checked

Done.


ee/backup/handler_s3.go, line 27 at r8 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

s3AccelerateHost is unused

Done.


ee/backup/handler_s3.go, line 28 at r8 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

const s3MinioChunkSize is unused

Done.


ee/backup/handler_s3.go, line 103 at r8 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

type progress is unused

Done.


ee/backup/writer.go, line 18 at r7 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

dgraphBackupSuffix is unused

Done.


worker/backup.go, line 40 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

return x.Errorf

Done.


worker/backup.go, line 50 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Instead of copying, why not pass pb.BackupRequest to backup package.

Done.


worker/backup.go, line 77 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

There's something not right about a function spawning a goroutine and just leaving them there. I find it easier to follow and a good practice when the func which spawns a goroutine also ensures that the goroutine reaches its conclusion.

This func backupDispatch, probably an apt name for what it's doing right now, should really do the dispatching, but also run through the results, returning a single error (not a chan).

We catch all the errors at the receiver in the order they finish and stop if we get the first error. the part missing (which i'm adding now) is a sync stop to stop any acting go routines.


worker/backup.go, line 132 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

There';s

Done.


x/file.go, line 44 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Can probably check? x.Check2(w.Write(b))

Done.


ee/backup/handler_file.go, line 35 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Can be V(2).

Done.


ee/backup/handler_file.go, line 47 at r5 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

If Sync fails, then just do fp.Close().

But, do return the error of Close, so remove the defer, and bring it inline after Sync.

Done.

if err != nil {
return x.Errorf("Error while looking for bucket: %s at host: %s. Error: %v",
h.bucket, uri.Host, err)
return err

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

unreachable code

Copy link
Contributor

@manishrjain manishrjain left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

:lgtm: The PR is good to go! Tested and all.

Reviewable status: 11 of 20 files reviewed, 23 unresolved discussions (waiting on @manishrjain, @golangcibot, and @srfrog)


ee/backup/backup.go, line 66 at r9 (raw file):

		return err
	}
	if err = w.cleanup(); err != nil {

Close is the typical expected fn call at the end, or Flush. Cleanup sounds strange here.

Copy link
Contributor Author

@srfrog srfrog left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewable status: 9 of 20 files reviewed, 23 unresolved discussions (waiting on @manishrjain and @golangcibot)


ee/backup/backup.go, line 66 at r9 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Close is the typical expected fn call at the end, or Flush. Cleanup sounds strange here.

Done.


ee/backup/s3_handler.go, line 87 at r10 (raw file):

Previously, golangcibot (Bot from GolangCI) wrote…

unreachable code

Done.

@srfrog srfrog merged commit bf0cfee into master Nov 8, 2018
@srfrog srfrog deleted the feature/roadmap-backups branch November 10, 2018 02:34
dna2github pushed a commit to dna2fork/dgraph that referenced this pull request Jul 19, 2019
* moved orchestrate to its own pkg within worker.
adding backup to worker and alpha.

* trying to get the handler and file writers working.

* added destination parameter. handler support to destination URI scheme. additional logging.

* file handler rename on same volume. added more comments and logging.

* changed worker to use stream pkg. updated protos for backup. fixed mislabelled request field.

* logging changes for debugging

* added some error checks, tweaked comments.

* moved stream pkg out of worker.
removed binary encoding package, using encode/binary with size delimiter.
fixed race and added posting list values to backup.
fixed issue with file handler that was breaking badger.
refactored backup process to be simpler.
added generic Status proto for any service response.
added minio and used it for S3 backup uploads.

* removed unused const. format fixes.

* Initial pass at simplifying things.

* cleaned up redundant code.
renamed s3handler.send to s3handler.upload and removed all buffering.
s3handler tests that bucket exists before working, we cant assume a region.
s3handler.Close blocks until the upload is complete.

* unused const

* missing space

* added progress monitoring. fixed issues found by CI

* Small fixes here and there.

* Rename handler files.

* Both S3 uploads and file writes are tested to work.

* renamed writer.cleapup to writer.close

* regenerated protos

* removed unneeded fallthrough
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants