Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I'm trying to use Amazon S3 service to store logs from my applications. Given /user/bin/s3cmd --help tells me what I need to know how send the files:

s3cmd --help
usage: s3cmd [options] COMMAND [parameters]

S3cmd is a tool for managing objects in Amazon S3 storage. It allows for
making and removing "buckets" and uploading, downloading and removing
"objects" from these buckets.

  -h, --help            show this help message and exit
  --configure           Invoke interactive (re)configuration tool.
  -c FILE, --config=FILE
                        Config file name. Defaults to
  --dump-config         Dump current configuration after parsing config files
                        and command line options and exit.
  -n, --dry-run         Only show what should be uploaded or downloaded but
                        don't actually do it. May still perform S3 requests to
                        get bucket listings and other information though (only
                        for file transfer commands)
  -e, --encrypt         Encrypt files before uploading to S3.
  --no-encrypt          Don't encrypt files.
  -f, --force           Force overwrite and other dangerous operations.
  --continue            Continue getting a partially downloaded file (only for
                        [get] command).
  --skip-existing       Skip over files that exist at the destination (only
                        for [get] and [sync] commands).
  -r, --recursive       Recursive upload, download or removal.
  --check-md5           Check MD5 sums when comparing files for [sync].
  --no-check-md5        Do not check MD5 sums when comparing files for [sync].
                        Only size will be compared. May significantly speed up
                        transfer but may also miss some changed files.
  -P, --acl-public      Store objects with ACL allowing read for anyone.
  --acl-private         Store objects with default ACL allowing access for you
                        Grant stated permission to a given amazon user.
                        Permission is one of: read, write, read_acp,
                        write_acp, full_control, all
                        Revoke stated permission for a given amazon user.
                        Permission is one of: read, write, read_acp, wr
                        ite_acp, full_control, all
  --delete-removed      Delete remote objects with no corresponding local file
  --no-delete-removed   Don't delete remote objects.
  -p, --preserve        Preserve filesystem attributes (mode, ownership,
                        timestamps). Default for [sync] command.
  --no-preserve         Don't store FS attributes
  --exclude=GLOB        Filenames and paths matching GLOB will be excluded
                        from sync
  --exclude-from=FILE   Read --exclude GLOBs from FILE
  --rexclude=REGEXP     Filenames and paths matching REGEXP (regular
                        expression) will be excluded from sync
  --rexclude-from=FILE  Read --rexclude REGEXPs from FILE
  --include=GLOB        Filenames and paths matching GLOB will be included
                        even if previously excluded by one of
                        --(r)exclude(-from) patterns
  --include-from=FILE   Read --include GLOBs from FILE
  --rinclude=REGEXP     Same as --include but uses REGEXP (regular expression)
                        instead of GLOB
  --rinclude-from=FILE  Read --rinclude REGEXPs from FILE
                        Datacentre to create bucket in. As of now the
                        datacenters are: US (default), EU, us-west-1, and ap-
  --reduced-redundancy, --rr
                        Store object with 'Reduced redundancy'. Lower per-GB
                        price. [put, cp, mv]
                        Target prefix for access logs (S3 URI) (for [cfmodify]
                        and [accesslog] commands)
  --no-access-logging   Disable access logging (for [cfmodify] and [accesslog]
  -m MIME/TYPE, --mime-type=MIME/TYPE
                        Default MIME-type to be set for objects stored.
  -M, --guess-mime-type
                        Guess MIME-type of files by their extension. Falls
                        back to default MIME-Type as specified by --mime-type
                        Add a given HTTP header to the upload request. Can be
                        used multiple times. For instance set 'Expires' or
                        'Cache-Control' headers (or both) using this options
                        if you like.
  --encoding=ENCODING   Override autodetected terminal and filesystem encoding
                        (character set). Autodetected: UTF-8
  --verbatim            Use the S3 name as given on the command line. No pre-
                        processing, encoding, etc. Use with caution!
  --list-md5            Include MD5 sums in bucket listings (only for 'ls'
  -H, --human-readable-sizes
                        Print sizes in human readable form (eg 1kB instead of
  --progress            Display progress meter (default on TTY).
  --no-progress         Don't display progress meter (default on non-TTY).
  --enable              Enable given CloudFront distribution (only for
                        [cfmodify] command)
  --disable             Enable given CloudFront distribution (only for
                        [cfmodify] command)
  --cf-add-cname=CNAME  Add given CNAME to a CloudFront distribution (only for
                        [cfcreate] and [cfmodify] commands)
                        Remove given CNAME from a CloudFront distribution
                        (only for [cfmodify] command)
  --cf-comment=COMMENT  Set COMMENT for a given CloudFront distribution (only
                        for [cfcreate] and [cfmodify] commands)
                        Set the default root object to return when no object
                        is specified in the URL. Use a relative path, i.e.
                        default/index.html instead of /default/index.html or
                        s3://bucket/default/index.html (only for [cfcreate]
                        and [cfmodify] commands)
  -v, --verbose         Enable verbose output.
  -d, --debug           Enable debug output.
  --version             Show s3cmd version (1.0.0) and exit.
  -F, --follow-symlinks
                        Follow symbolic links as if they are regular files

But it doesn't tell how check if the file were sent and remove the sent ones. Should I check via MD5 and delete locally by some shell script ?

share|improve this question
Use --check-md5 to determine if the files you uploaded are in sync. –  Ramhound May 9 '13 at 13:12
Add this parameter when I'm using the put operation ? Or after of it ? –  Valter Henrique May 9 '13 at 14:02
Its doubtful that it matters based on the description of --check-md5 try both with a sample file and post the results. –  Ramhound May 9 '13 at 15:21

2 Answers 2

up vote 1 down vote accepted

After some time I was able to develop a code in bash which check the md5sum from both, s3 and my local files and remove the local files that are already in amazon s3:


s3=`s3cmd ls --list-md5 -H s3://company-backup/company/"$datacenter"/"$hostname"/"$path"/`

s3_list=`echo "$s3"|awk {'print $4" "$5'} | sed 's= .*/= ='`

locally=`md5sum /"$path"/*.gz`;
locally_list=$(echo "$locally" | sed 's= .*/= =');
#echo "$locally_list";

for i in $locally_list
  #echo $i
  locally_hash=`echo $i|awk {'print $1'}`
  locally_file=`echo $i|awk {'print $2'}`

  for j in $s3_list
    s3_hash=$(echo $j|awk {'print $1'}); 
    s3_file=$(echo $j|awk {'print $2'});

    #to avoid empty file when have only hash from folder
    if [[ $s3_hash != "" ]] && [[ $s3_file != "" ]]; then 
      if [[ $s3_hash == $locally_hash ]] && [[ $s3_file == $locally_file ]]; then
        echo "### REMOVING ###";
        echo "$locally_file";
        #rm /"$path"/"$locally_file";
unset IFS
share|improve this answer

FWIW, I needed to do something similar and wrote the following bash script. What it does is:

  1. gets a list of files in a directory that are older than $MINUTES minutes old using find
  2. uses lsof to determine if the file is open (this may not be true if the file is say open by an editor)
  3. uses s3cmd to copy the file into an S3 bucket.
  4. compares MD5 sums on the remote file in S3 and the local one. If they check out, delete the local one.


TARGET_DIR="s3://AWSbucketname/subfolder/`hostname -s`/"

echo ""
echo "About to upload files in $LOCAL_DIR up to S3 folder:"
echo "    $TARGET_DIR"
echo "Then delete if MD5 sums line up."
echo "Starting in 5 seconds..."
sleep 5


# Throw the list of files that the find command gets into an array
while IFS= read -d $'\0' -r file ; do
    FILES=("${FILES[@]}" "$file")
done < <(find $LOCAL_DIR -name \*.wav -mmin +$MINUTES -print0)

# echo "${WAV_FILES[@]}"   # DEBUG

for local_file in "${WAV_FILES[@]}"
    # Check that the file in question is not open.
    # lsof returns non-zero return value for file not in use
    lsof "$local_file" 2>&1 > /dev/null
    if test $? -ne 0 ; then
        echo ""
        echo "$local_file isn't open. Copying to S3..."
        s3cmd -p put $local_file $TARGET_DIR
        # s3cmd -n put $local_file $TARGET_DIR # DEBUG - dry-run

        ## Now attempt to delete if the MD5 sums check out:

        md5sum_remote=`s3cmd info  "$TARGET_DIR$remote_file" | grep MD5 | awk '{print $3}'`
        md5sum_local=`md5sum $local_file | awk '{print $1}'`
        if [[ "$md5sum_remote" == "$md5sum_local" ]]; then
          echo "$remote_file MD5 sum checks out. Deleting..."
          rm $local_file
share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.