vladgh / docker_base_images Goto Github PK

Retry command every 10 seconds until it succeeds.

Use binary by default.

When restoring a backup using a prefix this function causes the prefix to get doubled so the aws s3 cp command fails.

get_archive(){
  if [[ -s $_restore_file ]]; then
    return
  else
    # Download the latest archive from S3
    set_up_s3
    _restore_file="$(get_latest_s3_archive)"

    if [[ -n "$_restore_file" ]]; then
      aws s3 cp "${_aws_s3_path}/${_restore_file}" "$_restore_file"
    fi
  fi
}

With a prefix of "myprefix" the script attempts to copy the backup from: "myprefix/myprefix/hourly..."

The issue is that aws_s3_path has the prefix at the end, and restore_file has the prefix at the start.

Hey, I'm using your backup image and it works great but I'm having an issue restoring.

When I call the restore command I get the error:

fatal error: Parameter validation failed: Invalid bucket name "s3:": Bucket name must match the regex "^[a-zA-Z0-9.\-_]{1,255}$"

If I echo out the bucket name it's trying to use I get:

s3://s3://[my-bucket]/[prefix]/[prefix]/hourly/2017-06-28_03-00-04.tar.xz

so it seem the s3:// gets doubled and the [prefix] gets doubled as well in the url construction.

Changing linke 167 in entrypoint.sh from

aws s3 cp "s3://${AWS_S3_PATH}/${RESTORE_FILE}" "$RESTORE_FILE"

to

aws s3 cp "s3://${AWS_S3_BUCKET}/${RESTORE_FILE}" "$RESTORE_FILE"

Seems to fix the path.

Example

docker run --rm -i \
  -e GPG_PASSPHRASE=myverystrongpassword \
  -v /host/path/to/restore:/restore \
  vladgh/backup restore < /path/to/host/restore_file.tar.xz.gpg

Refactor entrypoint.sh for a cleaner logic.

Server time seems to be the best option: https://docs.puppet.com/puppetdb/4.4/api/meta/v1/server-time.html

Your main:

main(){
  if [[ ! "$S3PATH" =~ s3:// ]]; then
    log 'No S3PATH specified' >&2; exit 1
  fi

  # Run initial sync
  sync_files "$S3PATH" "$DESTINATION"

  # Exit if argument is 'once'
  if [[ "${*:-}" == 'once' ]]; then exit; fi

  # Setup inotify or cron job
  if [[ -n "$WATCHDIR" ]]; then
    watch_directory
  elif [[ -n "$CRON_TIME" ]]; then
    run_cron
  fi
}

Note that the initial sync is run even if WATCHDIR is specified. I would assume that the behavior would be that if a WATCHDIR is specified, only that directory is synced to S3. I'm assuming there should be no sync from S3 -> destination in this case.

The healthcheck for the puppetserver image uses puppet to get the certificate paths. This adds an unnecessary burden to the CPU for every interval.
A simple hostname interpolation would be much better (ex: --cert "/etc/puppetlabs/puppet/ssl/certs/$(hostname).pem"), which should work just as well, but keep the load to a minimum.

Hi @vladgh.

Your dockerized minidlna really great! Thank you for it.

Is it possible to add support for more formats for dlna server? Like example, avi is not exposed by server to clients. Maybe in scope of this, if some additional codecs are missing too it could be added to image.

Thanks

Docker 17.06 allows specifying a secret location within the container

In the logs this looks like:
Completed 256.0 KiB/3.8 MiB (481.2 KiB/s) with 1 file(s) remaining#015Completed 512.0 KiB/3.8 MiB (869.8 KiB/s) with 1 file(s) remaining#015Completed 768.0 KiB/3.8 MiB (1.2 MiB/s) with 1 file(s) remaining #015Completed 1.0 MiB/3.8 MiB (1.5 MiB/s) with 1 file(s) remaining #015Completed 1.2 MiB/3.8 MiB (1.8 MiB/s) with 1 file(s) remaining #015Completed 1.5 MiB/3.8 MiB (2.1 MiB/s) with 1 file(s) remaining #015Completed 1.8 MiB/3.8 MiB (2.4 MiB/s) with 1 file(s) remaining #015Completed 2.0 MiB/3.8 MiB (2.6 MiB/s) with 1 file(s) remaining #015Completed 2.2 MiB/3.8 MiB (2.9 MiB/s) with 1 file(s) remaining #015Completed 2.5 MiB/3.8 MiB (3.1 MiB/s) with 1

Use SHA512 for hashing the password.

Backup script should avoid writing anything to disk, and use pipes instead (tar | gpg | s3).

For example GPG_RECIPIENT and GPG_KEY_URL will only work for the first time, not in subsequent cron jobs.

Useful for postrun hooks.