Commit Graph

73 Commits

Author SHA1 Message Date
Harshavardhana
51fa4f7fe3 Make PutObject a nop for an object which ends with "/" and size is '0' (#3603)
This helps majority of S3 compatible applications while not returning
an error upon directory create request.

Fixes #2965
2017-01-20 16:33:01 -08:00
Harshavardhana
926c75d0b5 api: Set appropriate content-type for success/error responses. (#3537)
Golang HTTP client automatically detects content-type but
for S3 clients this content-type might be incorrect or
might misbehave.

For example:

```
Content-Type: text/xml; charset=utf-8
```

Should be
```
Content-Type: application/xml
```

Allow this to be set properly.
2017-01-06 00:37:00 -08:00
Bala.FA
6d10f4c19a Adopt dsync interface changes and major cleanup on RPC server/client.
* Rename GenericArgs to AuthRPCArgs
* Rename GenericReply to AuthRPCReply
* Remove authConfig.loginMethod and add authConfig.ServiceName
* Rename loginServer to AuthRPCServer
* Rename RPCLoginArgs to LoginRPCArgs
* Rename RPCLoginReply to LoginRPCReply
* Version and RequestTime are added to LoginRPCArgs and verified by
  server side, not client side.
* Fix data race in lockMaintainence loop.
2017-01-02 20:57:42 +05:30
Harshavardhana
4daa0d2cee lock: Moving locking to handler layer. (#3381)
This is implemented so that the issues like in the
following flow don't affect the behavior of operation.

```
GetObjectInfo()
.... --> Time window for mutation (no lock held)
.... --> Time window for mutation (no lock held)
GetObject()
```

This happens when two simultaneous uploads are made
to the same object the object has returned wrong
info to the client.

Another classic example is "CopyObject" API itself
which reads from a source object and copies to
destination object.

Fixes #3370
Fixes #2912
2016-12-10 16:15:12 -08:00
Bala FA
bef0a50bc1 Cleanup and fixes (#3273)
* newRequestID() (previously generateUploadID()) returns string than byte array.
* Remove unclear comments and added appropriate comments.
* SHA-256, MD5 Hash functions return Hex/Base64 encoded string than byte array.
* Remove duplicate MD5 hasher functions.
* Rename listObjectsValidateArgs() into validateListObjectsArgs()
* Remove repeated auth check code in all bucket request handlers.
* Remove abbreviated names in bucket-metadata
* Avoid nested if in bucketPolicyMatchStatement()
* Use ioutil.ReadFile() instead of os.Open() and ioutil.ReadAll()
* Set crossDomainXML as constant.
2016-11-21 13:51:05 -08:00
Harshavardhana
398421b9f5 xl/bootup: Server bootup shouldn't return for missing buckets. (#3255)
Ref #3196
2016-11-14 15:45:00 -08:00
Aditya Manthramurthy
dd0698d14c Improve namespace lock API: (#3203)
- abstract out instrumentation information.
- use separate lockInstance type that encapsulates the nsMutex, volume,
  path and opsID as the frontend or top-level lock object.
2016-11-09 10:58:41 -08:00
Aditya Manthramurthy
85a5c358d8 Add bucket metadata state client/handler (Fixes #3022) (#3152)
- Adds an interface to update in-memory bucket metadata state called
  BucketMetaState - this interface has functions to:
     - update bucket notification configuration,
     - bucket listener configuration,
     - bucket policy configuration, and
     - send bucket event

- This interface is implemented by `localBMS` a type for manipulating
  local node in-memory bucket metadata, and by `remoteBMS` a type for
  manipulating remote node in-memory bucket metadata.

- The remote node interface, makes an RPC call, but the local node
  interface does not - it updates in-memory bucket state directly.

- Rename mkPeersFromEndpoints to makeS3Peers and refactored it.

- Use arrayslice instead of map in s3Peers struct

- `s3Peers.SendUpdate` now receives an arrayslice of peer indexes to
  send the request to, with a special nil value slice indicating that
  all peers should be sent the update.

- `s3Peers.SendUpdate` now returns an arrayslice of errors, representing
  errors from peers when sending an update. The array positions
  correspond to peer array s3Peers.peers

Improve globalS3Peers:

- Make isDistXL a global `globalIsDistXL` and remove from s3Peers

- Make globalS3Peers an array of (address, bucket-meta-state) pairs.

- Fix code and tests.
2016-11-07 12:09:24 -08:00
Aditya Manthramurthy
30dc11a931 No listener.json for single-node mode (Fixes #3052) (#3108)
In FS or single-node XL mode, there is no need to save listener
configuration to persistent storage. As there is only one server, if it
is restarted, any connected listenBucketAPI clients were disconnected
and will have to reconnect - so there is nothing to actually store.

This incidentally solves #3052 by avoiding the problem.
2016-10-26 20:13:00 -07:00
Aditya Manthramurthy
3977d6b7bd Lock bucket while modifying its metadata (Fixes #2979) (#3019)
- When modifying notification configuration
- When modifying listener configuration
- When modifying policy configuration

With this change we also stop early checking if the bucket exists, since
that uses a Read-lock and causes a deadlock due to the outer Write-lock.
2016-10-24 19:52:24 -07:00
Harshavardhana
83b364891d tests: Fix a potential race in ListenBucketNotificationHandler. (#3040) 2016-10-21 22:56:27 -07:00
Harshavardhana
ece559afe2 api: Do not use sqs for ListenBucketNotification. (#3023)
Add more tests. Fixes #3024
2016-10-21 01:25:17 -07:00
Aditya Manthramurthy
d02cb963d5 Fix listen-bucket (Fixes #2942) (#2949)
Don't close socket while re-initializing notify-listeners, as the rpc
client object is shared between notify-listeners and peer clients.

Also, improves SendRPC() readability by using GetPeerClient().
2016-10-16 20:52:10 -07:00
Mateusz Gajewski
73982c8cb6 Listen bucket notification for multiple prefixes/suffixes (#2911)
* Listen bucket notification for multiple prefixes/suffixes

* After review fixes
2016-10-12 11:02:15 -07:00
Aditya Manthramurthy
6199aa0707 Peer RPCs for bucket notifications (#2877)
* Implements a Peer RPC router that sends info to all Minio servers in the cluster.
* Bucket notifications are propagated to all nodes via this RPC router.
* Bucket listener configuration is persisted to separate object layer
  file (`listener.json`) and peer RPCs are used to communicate changes
  throughout the cluster.
* When events are generated, RPC calls to send them to other servers
  where bucket listeners may be connected is implemented.
* Some bucket notification tests are now disabled as they cannot work in
  the new design.
* Minor fix in `funcFromPC` to use `path.Join`
2016-10-12 01:03:50 -07:00
Krishna Srinivas
61a18ed48f sha256: Verify sha256 along with md5sum, signature is verified on the request early. (#2813) 2016-10-02 15:51:49 -07:00
Krishnan Parthasarathi
5fdd768903 Make addition of TopicConfig to globalEventNotifier go-routine safe (#2806) 2016-09-28 22:46:19 -07:00
Harshavardhana
ca3022d545 api: Change ListenBucketNotification with new API format. (#2791)
Take prefix, suffix and events as query params.
2016-09-27 13:17:43 -07:00
Harshavardhana
4917038f55 Move the ObjectAPI() resource to be beginning of each handlers.
We should return back proper errors so that the clients can
retry until server has been initialized.
2016-09-13 21:18:30 -07:00
Krishnan Parthasarathi
e55926e8cf distribute: Make server work with multiple remote disks
This change initializes rpc servers associated with disks that are
local. It makes object layer initialization on demand, namely on the
first request to the object layer.

Also adds lock RPC service vendorized minio/dsync
2016-09-13 21:18:30 -07:00
Harshavardhana
040af08473 server: Startup message now prints configured ARNs. (#2653)
Fixes #2628
2016-09-10 02:23:28 -07:00
Anis Elleuch
07506358ff Refactor Heal RPC and add Shutdown RPC (#2488) 2016-08-21 12:06:53 -07:00
Harshavardhana
bccf549463 server: Move all the top level files into cmd folder. (#2490)
This change brings a change which was done for the 'mc'
package to allow for clean repo and have a cleaner
github drop in experience.
2016-08-18 16:23:42 -07:00