Commit Graph

688 Commits

Author SHA1 Message Date
Harshavardhana dc0df3dc0e Breakaway from driver model, move cache into donut 2015-07-02 21:04:03 -07:00
Harshavardhana 72572d6c71 Remove some api server code bringing in new cleanup 2015-07-02 21:04:03 -07:00
Harshavardhana c2031ca066 Add server and control command 2015-07-02 21:04:03 -07:00
Frederick F. Kautz IV cfbc169034 Fixing API definitions 2015-07-02 13:14:21 -07:00
Harshavardhana 8f61d6b6be Move memory code out, add it as layer on top of existing donut code
Just like how http.Handlers can be overlayed on top of each other
with each implementing ServeHTTP().

drivers.Driver can be overlayed on top of each other in similar manner
which would implement the drivers.Driver interface.

   API <----> cache <----> donut <----> donut(format)
2015-06-30 10:09:12 -07:00
Harshavardhana ab6e16bb41 Add dummy driver for community to submit new drivers 2015-06-29 16:43:50 -07:00
Harshavardhana 12de98fb62 Rename memory driver as cache 2015-06-29 16:43:50 -07:00
Harshavardhana 2571342451 Filesystem goes the high road *again* 2015-06-29 16:43:42 -07:00
Harshavardhana f74d6138da IsValidBucket() is sufficient we don't need to verify for "." 2015-06-29 15:15:54 -07:00
Harshavardhana 3109909355 Handle couple of cases of OOM conditions, move caching to GetObject() rather than PutObject() 2015-06-29 12:28:50 -07:00
Harshavardhana d07d0c670a Return back proper errors in writeObjectData(), rename few functions 2015-06-29 11:46:35 -07:00
Harshavardhana 10c807f233 Put object on successful write returns full metadata, to avoid subsequent GetObjectMetadata() calls in driver 2015-06-29 11:15:46 -07:00
Harshavardhana 6921328b93 Avoid frivolous GetObjectMetadata() calls at driver level, return back all the information in donut ListObjects() 2015-06-29 11:14:58 -07:00
Harshavardhana d8f7896a43 Expand http server struct to store more values 2015-06-29 00:12:28 -07:00
Harshavardhana 42c0287943 Use errorChannels only for services not for drivers, reduce them to use simple functions 2015-06-28 23:59:47 -07:00
Harshavardhana 91e5f648cb Move to Set() not Append() due to large memory reference copy 2015-06-28 10:13:12 -07:00
Harshavardhana ac4d8fe478 Add free() method for proxyReader to aggressively de-allocate Read data, to handle certain out of memory conditions
There are still some more out there
2015-06-27 20:43:25 -07:00
Harshavardhana 05f8654e3d Add Append() method to trove cache for appending data to an existing key
This largely avoids a large buffer copy which would accumulate inside proxyReader{}

This patch also implements "initialize()" function to init and populate data
on all the existing buckets, avoiding the redundant ListBuckets() invoked by
every API call.
2015-06-27 20:25:24 -07:00
Harshavardhana 367772b988 Make sure to populate on disk data into memory upon first API requests 2015-06-27 18:25:21 -07:00
Harshavardhana 45a7eab804 An attempt to bring in memory layer into donut driver 2015-06-27 17:23:34 -07:00
Harshavardhana f3c25bcfc4 Keeping the lexical order same add optimizations, provide a comprehensive response from ListObjects() 2015-06-27 13:12:44 -07:00
Harshavardhana ae66ae42c4 Rename functions for their purpose 2015-06-27 12:39:11 -07:00
Harshavardhana 9a4680475f Object metadata was wrongly misconstrued to be mutable, handle it 2015-06-26 23:22:53 -07:00
Harshavardhana 3aa6d90c5e Avoid racy maps, read from disk on success return quickly. Many more optimizations 2015-06-26 19:49:37 -07:00
Harshavardhana 68974918ac Donut moves to typed metadata, removing the necessity for strconv, parsing and string converstions 2015-06-26 16:23:12 -07:00
Harshavardhana 767d3743ee Add new metadata definitions and use them 2015-06-26 13:34:09 -07:00
Harshavardhana 9958e34772 Import quick key value store from Minio Client for persistent state files, primarily for donut 2015-06-25 19:57:31 -07:00
Harshavardhana fb9adb5524 Donut cleanup another set
- Make sure to close all readers
- Fix errors in api_testsuite c.Assert(err, IsNil) should be done right after each function call
2015-06-25 18:54:34 -07:00
Harshavardhana eec66f195a Take all the ListObjects into bucket handlers
Earlier the listing would wait for all the objects to be processed
this is essentially very time consuming considering even for 100,000
files.
2015-06-25 18:04:29 -07:00
Harshavardhana 45e9d25931 Remove more bloated code - simplify 2015-06-25 13:02:08 -07:00
Harshavardhana 82dcbf262d Add simple locking for donut API for now - fixes #671 2015-06-25 11:29:11 -07:00
Harshavardhana 03b4d3b308 Go vet fixes for donut 2015-06-24 21:07:03 -07:00
Harshavardhana 57a2b53178 Removing further bloated code simplifying 2015-06-24 21:03:39 -07:00
Harshavardhana a2c205ff2e Use external package disk for donut. 2015-06-24 21:03:02 -07:00
Harshavardhana 841ff01412 Move disk into its own package, remove bloated code 2015-06-24 20:13:47 -07:00
Harshavardhana 1682c748ac Remove unnecessary interfaces from donut, cleanup 2015-06-24 19:43:38 -07:00
Harshavardhana 3498872467 Add sha256 and sha512 windows compatibility layer 2015-06-24 14:39:41 -07:00
Harshavardhana 77d35b87d4 Fix a bug on windows regarding blockSSE3 calculation 2015-06-24 14:24:33 -07:00
Harshavardhana f1410731db Add windows code for sha1 and crc32c 2015-06-24 14:16:32 -07:00
Harshavardhana ba2d3dea74 Fix wrong TMPFS listing in donut filesystem map 2015-06-23 23:16:06 -07:00
Harshavardhana e818bc7187 Trim iodine path properly, so that now errors have github.com/minio/minio prefixed 2015-06-23 13:36:54 -07:00
Harshavardhana 2fd52ca284 Add missing strongly typed errors for Donut 2015-06-23 11:54:44 -07:00
Harshavardhana 7a060110ff Fix BUILDDEPS paths for golang installation 2015-06-22 12:07:23 -07:00
Harshavardhana 3bf64f5669 Verify d.donut to be non-nil, usually happens when multiple go versions compilations are linked - possible cause for #659 2015-06-20 11:03:17 -07:00
Harshavardhana 641f07cecf Use filepath everywhere instead of path.{} functions for portability - fixes #656 2015-06-18 16:02:45 -07:00
Harshavardhana 573a6134b2 Remove redundant !ok for map 2015-06-17 22:36:46 -07:00
Harshavardhana e9a3fd677a Hold lock on GetGlobalStateKey() - fixes #652 2015-06-17 20:37:49 -07:00
Harshavardhana b793f53d48 Minior change filter() to filterObjects() 2015-06-17 20:35:44 -07:00
Harshavardhana d4e71beb54 Add time.Since on buckets created time to avoid deleting buckets spuriously during eviction of objects 2015-06-13 11:28:04 -07:00
Harshavardhana 636788997b Minor optimization avoiding one unnecessary copy 2015-06-11 14:23:05 -07:00
Harshavardhana 6d44274db6 Make sure to reply back ETag with quotes for s3 compliance 2015-06-09 01:13:02 -07:00
Harshavardhana a947767339 Remove unnecessary tlsHelpers, SSL works with domain certificate 2015-06-08 20:54:41 -07:00
Harshavardhana c6ecd94420 Enable TLS and use it 2015-06-08 20:10:59 -07:00
Harshavardhana 64b014369c Return back entity too large for createObject requests bigger than cache size 2015-06-08 17:10:07 -07:00
Harshavardhana ab1976bc14 Load from persisted multipart session, now multipart uploads can resume even after restarting minio fs server 2015-06-08 15:15:37 -07:00
Harshavardhana 2976f109df More multipart fixes, filter out parts and multipart metadata from listObjects 2015-06-08 14:08:36 -07:00
Harshavardhana e08d59be00 Make sure to delete uploadid's from active session file properly 2015-06-08 12:25:51 -07:00
Harshavardhana 0d3710cb9b Store in json file instead of gob for fs metadata 2015-06-08 12:08:01 -07:00
Harshavardhana 8efc842b59 Add delete handlers and reply back as 'NotImplemented' instead of 404 2015-06-08 11:06:06 -07:00
Harshavardhana 8a320c4447 Enhance logging handler to print in humanize form for ContentLength 2015-06-08 02:14:09 -07:00
Harshavardhana cefd740002 Put back the older conn_limit 2015-06-06 18:11:38 -07:00
Harshavardhana 371651bde6 Add rate limiter instead of connection limit for now 2015-06-06 18:04:55 -07:00
Harshavardhana 21916dc0a1 Fix a deadlock in trove when evicting objects when we reach the maxSize used 2015-06-06 16:02:52 -07:00
Harshavardhana 8b43655d9f Make connection limit configurable 2015-06-06 14:31:50 -07:00
Harshavardhana a344f5b34d Misc changes before capturing proper video
- Disable management UI command option
- Remove featureflags option, filesystem and memory already support multipart
- Print informative messages after starting minio server
2015-06-03 18:27:15 -07:00
Harshavardhana 097f70a3c5 Enable filesystem for API tests, this patch also implements ListMultipartUploads() 2015-06-01 20:57:13 -07:00
Harshavardhana bfb9209689 Skip files and weird directories from listBuckets(), allowing this so that minio fs mode can be used as ftp service 2015-06-01 16:43:43 -07:00
Harshavardhana d86ba5f336 Implement multipart support
- implemented NewMultipartUpload()
 - implemented CreateObjectPart()
 - implemented ListObjectParts()
 - implemented AbortMultipartUpload()
 - implemented CompleteMultipartUpload()
2015-06-01 16:25:11 -07:00
Harshavardhana c07d48e7e8 ErrorResponse from minio server needed proper xml translations 2015-06-01 12:19:27 -07:00
Harshavardhana a91e519692 Make caching a package trove, and use it inside memory driver 2015-05-31 18:41:56 -07:00
Anis Elleuch 7e5a26f42c Protect the exported method Delete() 2015-05-31 20:29:42 +01:00
Harshavardhana 82a0eac659 Add filesystem factory functions and add related cli options 2015-05-30 03:23:06 -07:00
Harshavardhana 0cc63706bb Resurrect back fileystem driver 2015-05-30 03:12:01 -07:00
Anis Elleuch d360d0fec3 Forbid the upload of files bigger than the memory backend capacity 2015-05-28 22:07:02 +01:00
Harshavardhana 554a244b47 Add donut OSX support 2015-05-26 02:14:34 -07:00
Harshavardhana e46660a966 Use http.TimeFormat to return, S3 expects us to return in GMT not UTC 2015-05-25 20:17:04 -07:00
Harshavardhana 19250296c6 Handle authorization header better 2015-05-24 20:59:53 -07:00
Harshavardhana 9666f2e5bf Relax minium object size limit, one can really upload a 0byte file
For example
~~~
$ mc ls https://s3.amazonaws.com/ferenginar/test.go
[2015-05-23 12:02:06 PDT]     0B test.go

$ curl -i -X HEAD https://s3.amazonaws.com/ferenginar/test.go
HTTP/1.1 200 OK
x-amz-id-2: ZMWLriPH+uQJ8IsaMHmuNOU/FXvdSq+s6O7ugUI9hZ695XJTNAZ1utKxh03w5Jcf
x-amz-request-id: F5A11F533B74DD8F
Date: Sat, 23 May 2015 19:19:29 GMT
Last-Modified: Sat, 23 May 2015 19:02:06 GMT
ETag: "d41d8cd98f00b204e9800998ecf8427e"
Accept-Ranges: bytes
Content-Type: binary/octet-stream
Content-Length: 0
Server: AmazonS3
~~~
2015-05-23 12:19:16 -07:00
Harshavardhana e11f9110b6 add missing validate access keys after being extracted from AuthHeader 2015-05-21 19:23:47 -07:00
Harshavardhana ad60b97d52 Fix some donut TODO's regarding bucket created date 2015-05-21 11:13:49 -07:00
Harshavardhana cd817d34df Finish last bits of api testsuite on server 2015-05-21 02:57:03 -07:00
Harshavardhana f00bfd38da Bump gocyclo back to 16 2015-05-19 18:32:50 -07:00
Harshavardhana 3d61ba5930 Fix duplicate map allocation 2015-05-18 13:14:01 -07:00
Harshavardhana e4c0d574dc Update totalnumber of multiparts inside CreateObjectPart(), also introduce support for UploadIDMarker and Prefix for listMultipartUploads 2015-05-18 12:24:36 -07:00
Harshavardhana 7ce47130fd HEAD request should have Content-Length for only successful response, there is no response body for errors, just header is sufficient - fixes #603
http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.13

"""  in the case of the HEAD method, the size of the entity-body that would have been sent had the request been a GET. """
2015-05-15 21:55:59 -07:00
Harshavardhana 317637fcb8 Remove redundant WriteHeader calls 2015-05-15 20:58:28 -07:00
Harshavardhana 5c5761ae16 Remove unnecessary GetObjectMetadata prefix argument 2015-05-15 19:25:17 -07:00
Harshavardhana 6275238a47 For multiparts use its own cache avoids deadlocks during object expiration
Using same object cache as regular objects for multiparts causes deadlocks during object expiration routine.

As a side affect this also fixes the bug of listing multiparts during regular 'listing' of objects.
2015-05-14 20:26:59 -07:00
Harshavardhana c41ebf7535 Fixing a race in NewMultipartUpload() session for simultaneous requests, observed at play.minio.io 2015-05-14 19:11:19 -07:00
Harshavardhana 8fb427a892 Add more tests for multipart now in driver api_testsuite 2015-05-14 17:28:39 -07:00
Harshavardhana dba7c94571 Add tests for abortMultipartUpload, listobjectparts, listmultipartuploads 2015-05-14 17:07:30 -07:00
Harshavardhana 1bd94ec6ab An attempt to implement ListMultipartUploads() 2015-05-14 16:22:47 -07:00
Harshavardhana 82b043005a Avoid one last memory copy 2015-05-14 11:26:55 -07:00
Harshavardhana 5498c90788 Add a new validateContentTypeHandler{}, verify Accept header earlier 2015-05-13 12:22:15 -07:00
Harshavardhana de56909d55 More scripts and code from Minio-io to minio 2015-05-11 16:43:38 -07:00
Harshavardhana 8d06504068 Change all minio-io path to minio 2015-05-11 16:23:10 -07:00
Harshavardhana e39de4f603 Fix typo introduced in previous commit 2015-05-11 14:24:31 -07:00
Harshavardhana f45635524f Remove SignatureV2 support, bring in SignatureV4 header only validation for now 2015-05-11 13:38:40 -07:00
Harshavardhana b48b7f51ef Use convenience functions for url.Values over looping through keys, fix a bug in listObjectParts() 2015-05-10 11:32:41 -07:00
Harshavardhana 14b5828505 Handle partNumberMarker with listObjectParts now and other fixes 2015-05-09 19:39:00 -07:00