burry
This is burry
, the BackUp & RecoveRY tool for cloud native infrastructure services. Use burry
to back up and restore
critical infrastructure base services such as ZooKeeper and etcd. More…
burry
currently supports the following infra services and storage targets:
ZooKeeper | etcd | Consul | |
---|---|---|---|
Amazon S3 | B/R | B/R | B/R |
Azure Storage | []/[] | []/[] | []/[] |
Google Storage | B/R | B/R | B/R |
Local | B/R | B/R | B/R |
Minio* | B/R | B/R | B/R |
TTY** | B/- | B/- | B/- |
B ... backups supported
R ... restores supported
- ... not applicable
[] ... not yet implemented
*) Minio can be either on-premises or in the cloud, but always self-hosted. See also https://www.minio.io
**) TTY effectively means it's not stored at all but rather dumped on the screen; useful for debugging, though.
Note:
burry
is WIP, please use with care- if you want to learn more about the design (goals/assumptions), check out the background notes
- if you want to hack (rather than simply use)
burry
, check out the development and testing notes
Contents:
- Install
- Use
- Backups
- Example: Screen dump of local ZooKeeper content
- Example: Back up etcd to local storage
- Example: Back up DC/OS system ZooKeeper to Amazon S3
- Example: Back up etcd to Minio
- Restores
- Example: Restore etcd from local storage
- Example: Restore Consul from Minio
- Backups
- Release history
- Background
- Development and testing
Install
Currently, only 'build from source' install is available (note: replace GOOS=linux
with your platform):
$ go get github.com/mhausenblas/burry.sh
$ GOOS=linux go build
$ mv burry.sh burry
$ godoc -http=":6060" &
$ open http://localhost:6060/pkg/github.com/mhausenblas/burry.sh/
See also GoDocs.
Use
The general usage is:
$ burry --help
Usage: burry [args]
Arguments:
-b, --burryfest
Create a burry manifest file .burryfest in the current directory.
The manifest file captures the current command line parameters for re-use in subsequent operations.
-c, --credentials string
The credentials to use in format STORAGE_TARGET_ENDPOINT,KEY1=VAL1,...KEYn=VALn.
Example: s3.amazonaws.com,ACCESS_KEY_ID=...,SECRET_ACCESS_KEY=...
-e, --endpoint string
The infra service HTTP API endpoint to use.
Example: localhost:8181 for Exhibitor
-i, --isvc string
The type of infra service to back up or restore.
Supported values are [etcd zk] (default "zk")
-o, --operation string
The operation to carry out.
Supported values are [backup restore] (default "backup")
-s, --snapshot string
The ID of the snapshot.
Example: 1483193387
-t, --target string
The storage target to use.
Supported values are [local minio s3 tty] (default "tty")
-v, --version
Display version information and exit.
Note: If you want to re-use your command line parameters, use the --burryfest
or -b
argument: this creates a manifest file .burryfest
in the current directory, capturing all your settings. If a manifest .burryfest
exists in the current directory subsequent invocations use this and hence you can simply execute burry
, without any parameters. Remove the .burryfest
file in the current directory to reset stored command line parameters.
An example of a burry manifest file looks like:
{
"svc": "etcd",
"svc-endpoint": "etcd.mesos:1026",
"target": "local",
"credentials": {
"target-endpoint": "",
"params": []
}
}
Note that for every storage target other than tty
a metadata file .burrymeta
in the (timestamped) archive file will be created, something like:
{
"snapshot-date": "2016-12-31T14:52:42Z",
"svc": "zk",
"svc-endpoint": "leader.mesos:2181",
"target": "s3",
"target-endpoint": "s3.amazonaws.com"
}
Backups
In general, since --operation backup
is the default, the only required parameter for a backup operation is the --endpoint
. That is, you'll have to provide the HTTP API of the ZooKeeper or etcd you want to back up:
$ burry --endpoint IP:PORT (--isvc etcd|zk) (--target tty|local|s3) (--credentials STORAGE_TARGET_ENDPOINT,KEY1=VAL1,...,KEYn=VALn)
Some concrete examples follow now.
Screen dump of local ZooKeeper content
To dump the content of a locally running ZK onto the screen, do the following:
# launching ZK:
$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
9ae41a9a02f8 mbabineau/zookeeper-exhibitor:latest "bash -ex /opt/exhibi" 2 days ago Up 2 days 0.0.0.0:2181->2181/tcp, 0.0.0.0:2888->2888/tcp, 0.0.0.0:3888->3888/tcp, 0.0.0.0:8181->8181/tcp amazing_kilby
# dump to screen:
$ DEBUG=true ./burry --endpoint localhost:2181
INFO[0000] Using burryfest /home/core/.burryfest func=init
INFO[0000] My config: {InfraService:zk Endpoint:localhost:2181 StorageTarget:tty Creds:{StorageTargetEndpoint: Params:[]}} func=init
INFO[0000] /zookeeper/quota: func=reapsimple
INFO[0000] Operation successfully completed. func=main
See the development and testing notes for the test setup.
Back up etcd to local storage
To back up the content of an etcd running in a (DC/OS) cluster to local storage, do:
# create the backup:
$ ./burry --endpoint etcd.mesos:1026 --isvc etcd --target local
INFO[0000] My config: {InfraService:etcd Endpoint:etcd.mesos:1026 StorageTarget:local Creds:{StorageTargetEndpoint: Params:[]}} func=init
INFO[0000] Operation successfully completed. The snapshot ID is: 1483194168 func=main
# check for the archive:
$ ls -al 1483194168.zip
-rw-r--r--@ 1 mhausenblas staff 750 31 Dec 14:22 1483194168.zip
# explore the archive:
$ unzip 1483194168.zip && cat 1483194168/.burrymeta | jq .
{
"snapshot-date": "2016-12-31T14:22:48Z",
"svc": "etcd",
"svc-endpoint": "etcd.mesos:1026",
"target": "local",
"target-endpoint": "/tmp"
}
See the development and testing notes for the test setup.
Back up DC/OS system ZooKeeper to Amazon S3
To back up the content of the DC/OS system ZooKeeper (supervised by Exhibitor), do the following:
# let's first do a dry run:
$ ./burry --endpoint leader.mesos:2181
INFO[0000] My config: {InfraService:zk Endpoint:leader.mesos:2181 StorageTarget:tty Creds:{StorageTargetEndpoint: Params:[]}} func=init
INFO[0006] Operation successfully completed. func=main
# back up into Amazon S3:
$ ./burry --endpoint leader.mesos:2181 --target s3 --credentials s3.amazonaws.com,ACCESS_KEY_ID=***,SECRET_ACCESS_KEY=***
INFO[0000] My config: {InfraService:zk Endpoint:leader.mesos:2181 StorageTarget:s3 Creds:{InfraServiceEndpoint:s3.amazonaws.com Params:[{Key:ACCESS_KEY_ID Value:***} {Key:SECRET_ACCESS_KEY Value:***}]}}} func=init
INFO[0008] Successfully stored zk-backup-1483166506/latest.zip (45464 Bytes) in S3 compatible remote storage s3.amazonaws.com func=remoteS3
INFO[0008] Operation successfully completed. The snapshot ID is: 1483166506 func=main
See the development and testing notes for the test setup. Note: in order to back up to Google Storage rather than to Amazon S3, use --credentials storage.googleapis.com,ACCESS_KEY_ID=***,SECRET_ACCESS_KEY=***
in above command. Make sure that you have Google Storage as your default project and Interoperability enabled; see also settings in console.cloud.google.com/storage/settings.
Back up etcd to Minio
To back up the content of an etcd running in a (DC/OS) cluster to Minio, do:
$ ./burry --endpoint etcd.mesos:1026 --isvc etcd --credentials play.minio.io:9000,ACCESS_KEY_ID=Q3AM3UQ867SPQQA43P2F,SECRET_ACCESS_KEY=zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG --target s3
INFO[0000] My config: {InfraService:etcd Endpoint:etcd.mesos:1026 StorageTarget:s3 Credentials:} func=init
INFO[0001] Successfully stored etcd-backup-1483173687/latest.zip (674 Bytes) in S3 compatible remote storage play.minio.io:9000 func=remoteS3
INFO[0001] Operation successfully completed. The snapshot ID is: 1483173687 func=main
See the development and testing notes for the test setup. Note: the credentials used above are from the public Minio playground.
Restores
For restores you MUST set --operation restore
(or: -o restore
) as well as provide a snapshot ID with --snapshot/-s
. Note also that you CAN NOT restore from screen, that is, --target/-t tty
is an invalid choice:
$ burry --operation restore --target local|s3 --snapshot sID (--isvc etcd|zk) (--credentials STORAGE_TARGET_ENDPOINT,KEY1=VAL1,...,KEYn=VALn)
Restore etcd from local storage
In the following, we first create a local backup of an etcd cluster, then simulate failure by deleting a key and then restore it:
# let's first back up etcd:
$ ./burry -e etcd.mesos:1026 -i etcd -t local -b
INFO[0000] Selected operation: BACKUP func=init
INFO[0000] My config: {InfraService:etcd Endpoint:10.0.1.139:1026 StorageTarget:local Creds:{StorageTargetEndpoint: Params:[]}} func=init
INFO[0000] Operation successfully completed. The snapshot ID is: 1483383204 func=main
# now, let's destroy a key:
$ curl etcd.mesos:1026/v2/keys/foo -XDELETE
{"action":"delete","node":{"key":"/foo","modifiedIndex":16,"createdIndex":15},"prevNode":{"key":"/foo","value":"bar","modifiedIndex":15,"createdIndex":15}}
# restore it from the local backup:
$ ./burry -o restore -e etcd.mesos:1026 -i etcd -t local -s 1483383204
INFO[0000] Using burryfest /tmp/.burryfest func=init
INFO[0000] Selected operation: RESTORE func=init
INFO[0000] My config: {InfraService:etcd Endpoint:10.0.1.139:1026 StorageTarget:local Creds:{StorageTargetEndpoint: Params:[]}} func=init
INFO[0000] Restored /foo func=visitETCDReverse
INFO[0000] Operation successfully completed. Restored 1 items from snapshot 1483383204 func=main
# ... and we're back to normal:
$ curl 10.0.1.139:1026/v2/keys/foo
{"action":"get","node":{"key":"/foo","value":"bar","modifiedIndex":17,"createdIndex":17}}
See the development and testing notes for the test setup.
Restore Consul from Minio
In the following, we first create a backup of an Consul K/V store in Minio, then simulate failure by deleting a key and then restore it:
# let's first back up the Consul K/V store to Minio:
$ ./burry -e jump:8500 -i consul -t s3 -c play.minio.io:9000,ACCESS_KEY_ID=Q3AM3UQ867SPQQA43P2F,SECRET_ACCESS_KEY=zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG
INFO[0000] Selected operation: BACKUP func=main
INFO[0000] My config: {InfraService:consul Endpoint:jump:8500 StorageTarget:s3 Creds:{StorageTargetEndpoint:play.minio.io:9000 Params:[{Key:ACCESS_KEY_ID Value:Q3AM3UQ867SPQQA43P2F} {Key:SECRET_ACCESS_KEY Value:zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG}]}} func=main
INFO[0000] Operation successfully completed. The snapshot ID is: 1483448835 func=main
# now, let's destroy a key
$ curl jump:8500/v1/kv/foo -XDELETE
# restore it from the local backup:
$ ./burry-o restore -e jump:8500 -i consul -t s3 -s 1483448835 -c play.minio.io:9000,ACCESS_KEY_ID=Q3AM3UQ867SPQQA43P2F,SECRET_ACCESS_KEY=zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG
INFO[0000] Selected operation: RESTORE func=main
INFO[0000] My config: {InfraService:consul Endpoint:jump:8500 StorageTarget:s3 Creds:{StorageTargetEndpoint:play.minio.io:9000 Params:[{Key:ACCESS_KEY_ID Value:Q3AM3UQ867SPQQA43P2F} {Key:SECRET_ACCESS_KEY Value:zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG}]}} func=main
INFO[0000] Restored foo func=visitCONSULReverse
INFO[0000] Restored hi func=visitCONSULReverse
INFO[0000] Operation successfully completed. Restored 2 items from snapshot 1483448835 func=main
# ... and we're back to normal:
$ curl jump:8500/v1/kv/foo?raw
bar
See the development and testing notes for the test setup. Note: the credentials used above are from the public Minio playground.
Release history
- v0.4.0: support for backing up to and restoring from Google Storage
- v0.3.0: support for backing up and restoring Consul from local storage and S3/Minio
- v0.2.0: support for restoring ZK and etcd from local storage and S3/Minio
- v0.1.0: support for backing up ZK and etcd to screen, local storage and S3/Minio