Files
aptly/cmd/mirror_update.go
T
Andrey Smirnov bc7903f86e Rework mirror update (download packages) implementation
`PackageDownloadTask` is just a reference to file now. Whole process
was rewritten to follow pattern: download to temp location inside the pool,
verify/update checksums, import into pool as final step.

This removes a lot of edge cases when aptly internal state might be broken
if updating from rogue mirror.

Also this changes whole memory model: package list/files are kept in memory
now during the duration of `mirror update` command and saved to disk
only in the end.
2017-04-26 23:17:03 +03:00

267 lines
6.8 KiB
Go

package cmd
import (
"fmt"
"os"
"os/signal"
"strings"
"sync"
"github.com/smira/aptly/aptly"
"github.com/smira/aptly/deb"
"github.com/smira/aptly/query"
"github.com/smira/aptly/utils"
"github.com/smira/commander"
"github.com/smira/flag"
)
func aptlyMirrorUpdate(cmd *commander.Command, args []string) error {
var err error
if len(args) != 1 {
cmd.Usage()
return commander.ErrCommandError
}
name := args[0]
repo, err := context.CollectionFactory().RemoteRepoCollection().ByName(name)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
err = context.CollectionFactory().RemoteRepoCollection().LoadComplete(repo)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
force := context.Flags().Lookup("force").Value.Get().(bool)
if !force {
err = repo.CheckLock()
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
}
ignoreMismatch := context.Flags().Lookup("ignore-checksums").Value.Get().(bool)
maxTries := context.Flags().Lookup("max-tries").Value.Get().(int)
verifier, err := getVerifier(context.Flags())
if err != nil {
return fmt.Errorf("unable to initialize GPG verifier: %s", err)
}
err = repo.Fetch(context.Downloader(), verifier)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
context.Progress().Printf("Downloading & parsing package files...\n")
err = repo.DownloadPackageIndexes(context.Progress(), context.Downloader(), context.CollectionFactory(), ignoreMismatch, maxTries)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
if repo.Filter != "" {
context.Progress().Printf("Applying filter...\n")
var filterQuery deb.PackageQuery
filterQuery, err = query.Parse(repo.Filter)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
var oldLen, newLen int
oldLen, newLen, err = repo.ApplyFilter(context.DependencyOptions(), filterQuery, context.Progress())
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
context.Progress().Printf("Packages filtered: %d -> %d.\n", oldLen, newLen)
}
var (
downloadSize int64
queue []deb.PackageDownloadTask
)
skipExistingPackages := context.Flags().Lookup("skip-existing-packages").Value.Get().(bool)
context.Progress().Printf("Building download queue...\n")
queue, downloadSize, err = repo.BuildDownloadQueue(context.PackagePool(), skipExistingPackages)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
defer func() {
// on any interruption, unlock the mirror
err = context.ReOpenDatabase()
if err == nil {
repo.MarkAsIdle()
context.CollectionFactory().RemoteRepoCollection().Update(repo)
}
}()
repo.MarkAsUpdating()
err = context.CollectionFactory().RemoteRepoCollection().Update(repo)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
err = context.CloseDatabase()
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
// Catch ^C
sigch := make(chan os.Signal)
signal.Notify(sigch, os.Interrupt)
defer signal.Stop(sigch)
abort := make(chan struct{})
go func() {
<-sigch
signal.Stop(sigch)
close(abort)
}()
count := len(queue)
context.Progress().Printf("Download queue: %d items (%s)\n", count, utils.HumanBytes(downloadSize))
// Download from the queue
context.Progress().InitBar(downloadSize, true)
downloadQueue := make(chan deb.PackageDownloadTask)
var (
errors []string
errLock sync.Mutex
)
pushError := func(err error) {
errLock.Lock()
errors = append(errors, err.Error())
errLock.Unlock()
}
go func() {
for _, task := range queue {
select {
case downloadQueue <- task:
case <-abort:
return
}
}
queue = nil
close(downloadQueue)
}()
var wg sync.WaitGroup
for i := 0; i < context.Config().DownloadConcurrency; i++ {
wg.Add(1)
go func() {
defer wg.Done()
for {
select {
case task, ok := <-downloadQueue:
if !ok {
return
}
var e error
// provision download location
tempDownPath, e := context.PackagePool().(aptly.LocalPackagePool).GenerateTempPath(task.File.Filename)
if e != nil {
pushError(e)
continue
}
// download file...
e = context.Downloader().DownloadWithChecksum(
repo.PackageURL(task.File.DownloadURL()).String(),
tempDownPath,
&task.File.Checksums,
ignoreMismatch,
maxTries)
if e != nil {
pushError(e)
continue
}
// and import it back to the pool
task.File.PoolPath, e = context.PackagePool().Import(tempDownPath, task.File.Filename, &task.File.Checksums, true)
if e != nil {
pushError(e)
continue
}
// update "attached" files if any
for _, additionalTask := range task.Additional {
additionalTask.File.PoolPath = task.File.PoolPath
additionalTask.File.Checksums = task.File.Checksums
}
case <-abort:
return
}
}
}()
}
// Wait for all downloads to finish
wg.Wait()
select {
case <-abort:
return fmt.Errorf("unable to update: interrupted")
default:
}
context.Progress().ShutdownBar()
if len(errors) > 0 {
return fmt.Errorf("unable to update: download errors:\n %s", strings.Join(errors, "\n "))
}
err = context.ReOpenDatabase()
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
repo.FinalizeDownload(context.CollectionFactory())
err = context.CollectionFactory().RemoteRepoCollection().Update(repo)
if err != nil {
return fmt.Errorf("unable to update: %s", err)
}
context.Progress().Printf("\nMirror `%s` has been successfully updated.\n", repo.Name)
return err
}
func makeCmdMirrorUpdate() *commander.Command {
cmd := &commander.Command{
Run: aptlyMirrorUpdate,
UsageLine: "update <name>",
Short: "update mirror",
Long: `
Updates remote mirror (downloads package files and meta information). When mirror is created,
this command should be run for the first time to fetch mirror contents. This command can be
run multiple times to get updated repository contents. If interrupted, command can be safely restarted.
Example:
$ aptly mirror update wheezy-main
`,
Flag: *flag.NewFlagSet("aptly-mirror-update", flag.ExitOnError),
}
cmd.Flag.Bool("force", false, "force update mirror even if it is locked by another process")
cmd.Flag.Bool("ignore-checksums", false, "ignore checksum mismatches while downloading package files and metadata")
cmd.Flag.Bool("ignore-signatures", false, "disable verification of Release file signatures")
cmd.Flag.Bool("skip-existing-packages", false, "do not check file existence for packages listed in the internal database of the mirror")
cmd.Flag.Int64("download-limit", 0, "limit download speed (kbytes/sec)")
cmd.Flag.Int("max-tries", 1, "max download tries till process fails with download error")
cmd.Flag.Var(&keyRingsFlag{}, "keyring", "gpg keyring to use when verifying Release file (could be specified multiple times)")
return cmd
}