The Swift client
The Python Swift client is a command line tool for using object storage systems such as Allas. If you use Allas on Puhti or Mahti, all required packages and software are already installed.
Open a connection to Allas:The allas-conf command above asks for your CSC password (the same that you use to login to CSC servers). It lists your projects in Allas and asks you to define the project that will be used. allas-conf generates and authenticates the connection to the selected project in Allas. The authentication information is stored in the shell variables OS_AUTH_TOKEN and OS_STORAGE_URL that are valid for up to eight hours. However, you can refresh the authentication at any time by running allas-conf again. The environment variables are available only for that login session. If you open another session, you need to authenticate again to access Allas.
This chapter includes instructions for the following operations:
Swift command | Function |
---|---|
post | Create a bucket |
upload | Upload an object |
list | List objects and buckets |
download | Download objects and buckets |
copy | Move object |
delete | Remove objects or bucket |
download --all | Download whole project |
delete --all | Remove whole project |
md5sum | Get the checksum |
stat | View metadata |
stat --meta | Add metadata |
tempurl | Create a temporary URL |
post -r, -w, --read-acl | Manage access rights |
upload --use-slo | Upload files larger than 5 GB |
You can also install and use Swift outside the CSC computing environment. Make sure that your Swift version is not outdated, since the older Swift versions might not work with Allas.
Create buckets and upload objects
Create a new empty bucket:
Create a new bucket and add a file to it:
Add a file to an existing bucket:
Note: This might trigger the warning "409 Conflict: BucketAlreadyExists", but that does not necessarily mean that the upload has failed. If the next line displays the file name, the file was successfully uploaded to the existing bucket.$ swift upload my_fishbucket my_fish.jpg
Warning: failed to create container 'my_fishbucket': 409 Conflict: BucketAlreadyExists
my_fish.jpg
List objects and buckets
List all buckets belonging to a project:
List the content of a bucket:Download objects and buckets
Download an object:
If you want to rename the object as you download it, you can include -o new_name in the end of the command: Download an entire bucket:Move objects
You can copy data from one bucket to another using the command swift copy
. The command below copies file.txt from bucket1 to bucket2.
$ swift copy --destination /other_bucket my_bigfishes bigfish.jpg
created container other_bucket
my_bigfishes/bigfish.jpg copied to /other_bucket/bigfish.jpg
$ swift list other_bucket
bigfish.jpg
other_file.txt
Rename a file while copying it:
$ swift copy --destination /new_bucket/newname.jpg my_fishbucket my_fish.jpg
created container new_bucket
my_fishbucket/my_fish.jpg copied to /new_bucket/newname.jpg
For further information about the command swift copy, see the OpenStack documentation.
Remove objects and buckets
Remove objects and buckets using the command swift delete
:
Unlike with the web client and s3cmd, with Swift, you can delete an entire bucket at once:
For example:$ swift delete old_fishbucket
old_fish.png
useless_salmon.jpg
too_tiny_bass.jpg
$ swift list old_fishbucket
Container u'old_fishbucket' not found
Download or delete projects
Download the entire project:
Delete the entire project:
Please note: Be careful with this command since it deletes the entire content of the project. Before using this command, make sure you do not need the data anymore or that you have a copy of the data.Pseudo folders and checksums
In case you want to observe whether an object has changed, use checksum with the command md5sum
.
Pseudo folders can be handled by adding the name of the pseudo folder in front of the file name: my_pseudo_folder_name/my_file
Create a pseudo folder named pictures in the bucket my_bigfishes and add the object bass.png to it:
The example below uploads a file called salmon.jpg to the pseudo folder called fishes inside the bucket my_fishbucket. The file is then downloaded.
$ md5sum salmon.jpg
22e44aa2b856e4df892b43c63d15138a salmon.jpg
$ swift upload my_fishbucket/fishes salmon.jpg
fishes/salmon.jpg
$ swift list my_fishbucket
fishes/salmon.jpg
my_fish.jpg
$ swift download my_fishbucket fishes/salmon.jpg -o my_renamed_salmon.jpg
fishes/salmon.jpg [auth 0.664s, headers 0.925s, total 0.969s, 3.605 MB/s]
$ md5sum my_renamed_salmon.jpg
22e44aa2b856e4df892b43c63d15138a my_renamed_salmon.jpg
Managing metadata
Define metadata for an object:
Display details about a bucket:
$ swift stat my_fishbucket
Account: AUTH_$PROJECT_UUID
Container: my_fishbucket
Objects: 4
Bytes: 2162342
Read ACL:
Write ACL:
Sync To:
Sync Key:
Accept-Ranges: bytes
X-Trans-Id: txUUID-cpouta-production-kaj
X-Storage-Policy: default-placement
X-Container-Bytes-Used-Actual: 1167360
X-Timestamp: 1516776076.95812
Set a bucket as read-only to the world (make the content visible at the URL: a3s.fi/bucket_name/object_name) instead of the default (private to the project):
Find more information about access management in the section Giving another project read and write access to a bucket.
More details about a file:
$ swift stat my_fishbucket fishes/salmon.jpg
Account: AUTH_$PROJECT_ID
Container: my_fishbucket
Object: fishes/salmon.jpg
Content Type: image/jpeg
Content Length: 63220
Last Modified: Wed, 24 Jan 2018 10:17:03 GMT
ETag: a38f8db198e3fea43c83c465ffb0283b
Meta S3Cmd-Attrs: atime:1516788402/ctime:1513681753/gid:$LOCALGID/gname:$LOCALGROUP/md5:a38f8db198e3fea43c83c465ffb0283b/mode:33188/mtime:1513681747/uid:$LOCALUID/uname:$LOCALUSER
Accept-Ranges: bytes
X-Timestamp: 1516789023.84380
X-Trans-Id: tx0000000000000000001d6-q-q-cpouta-production-kaj
Note that the above file was uploaded with the s3cmd client, and therefore there is the additional metadata S3Cmd-Attrs compared to a file uploaded with Swift or S3. ETag is the hash when viewing the file details in the Pouta dashboard.
Giving another project read and write access to a bucket
Give the project project1 read rights to the bucket my_fishbucket:
Write access can be given similarly by replacing -r (read) with -w (write):
The character * after the project name defines that all project members in the project gain the rights.
Alternatively, you can give read and write access only to certain members of another project:
swift post my_fishbucket -r "project2:member1"
swift post my_fishbucket -w \
"project3:member1,project3:member2,project5:member1,project6:*"
Please note: If you have granted access for specific projects, making the shared project public and private again will remove any previous access permissions.
In case you allow -w access for another project, the members of the other project can upload files to your bucket and remove your files. However, you do not have access to the uploaded files until either you or the sender shares the bucket with your project:
For example:
Alternatively, you can set the project public and then access the file.
Files larger than 5 GB
Swift has a single-object size limit of 5 GiB. In order to upload files larger than this, you must create a large object that consists of smaller segments. To achieve this, you can use Swift to upload a so-called Static Large Object (SLO).
Try to upload a large file:
$ md5sum /tmp/6GB.zero
9e6a77a2d5650b2e2a710a08e9e61a81 /tmp/6GB.zero
$ stat /tmp/6GB.zero
File: '/tmp/6GB.zero'
Size: 6424625152 Blocks: 12548104 IO Block: 4096 regular file
...
$ swift upload my_bigfishes /tmp/6GB.zero
Object PUT failed: https://a3s.fi:443/swift/v1/my_bigfishes/tmp/6GB.zero 400 Bad Request EntityTooLarge
It fails with the message EntityTooLarge
, so instead:
$ swift upload my_bigfishes --use-slo --segment-size 1G /tmp/6GB.zero
tmp/6GB.zero segment 3
tmp/6GB.zero segment 5
tmp/6GB.zero segment 1
tmp/6GB.zero segment 0
tmp/6GB.zero segment 4
tmp/6GB.zero segment 2
tmp/6GB.zero
This creates a new bucket:
In this case the target bucket (my_bigfishes) contains just a front object that contains information what segments, stored in the segments bucket (my_bigfishes_segments) make the stored file. Operations performed to the front object are automatically reflected to the segments. Normally users don't need to operate with the segments buckets at all and objects inside these buckets should not be deleted or modified.
Download the entire 6GB.zero: