• Ability to request certain document fields from search results, by inclusion or exclusion:

{
    "q": "running",
    "fields": {"title": false, "price": false}
}

• Fixed a minor bug where facet sizes was not being respected in response. Only the max size was being honored

• Expose basic metrics on GRPC calls on /metics in prometheus format.
• Introduce make run_full, which will also start prometheus and grafana on top of tigris_server and fdb, where the metrics are visible.

• Add docker file for building monocontainer development image.
• Used docker compose plugin for local builds
• Makefile improvements:
  • Only rebuild and restart changed containers
  • Cache tests results (only run changed tests) (all caches, volumes still can be clean and all containers restart by make clean)

/search interface with stubbed response.

Request

{
    "q": "tunnel",
    "search_fields": [
        "first_name",
        "last_name"
    ],
    "filter": {
        "last_name": "Steve"
    },
    "facet": {
        "facet stat": 0
    },
    "sort": [
        {
            "salary": "$asc"
        }
    ],
    "fields": [
        "employment",
        "history"
    ]
}

Response

{
    "result": {
        "hits": [
            {
                "meta": {}
            }
        ],
        "facets": {
            "myField": {
                "counts": [
                    {
                        "count": 32,
                        "value": "adidas"
                    }
                ],
                "stats": {
                    "avg": 40,
                    "count": 50
                }
            }
        },
        "meta": {
            "found": 1234,
            "page": {
                "current": 2,
                "per_page": 10
            }
        }
    }
}

Fix installing dependencies for x86_64 macOS:

• wget is not available on macOS by default, use curl instead
• mktemp has different parameters
• SHA512 is calculated by different command

The modified version works on a fresh install of macOS.

Describe the bug For tests, I run several requests to create databases in parallel. Sometimes the error for database creation is nil however database is absent. Sometimes the error does appear.

To Reproduce

1. put files in one folder, run go mod tidy
2. run go test -count=1 ./...

Expected behavior Expected 3 databases to be present:

skydancer:~/tigriscli$ tigris list databases
testthree
testtwo
testone

Screenshots Instead, the error appears:

skydancer:~/tigriscli$ go test -count=1 ./... 
--- FAIL: TestThree (0.01s)
    tigriscli_test.go:37: recreate db testthree
    tigriscli_test.go:40: 
        	Error Trace:	tigriscli_test.go:40
        	            				tigriscli_test.go:25
        	Error:      	Received unexpected error:
        	            	database already exist
        	Test:       	TestThree
FAIL
FAIL	tigriscli	0.054s
FAIL

and the databases list if not full,

skydancer:~/tigriscli$ tigris list databases
testthree
testtwo

The name of the absent database testone is not the testthree that appeared in the error message.

Desktop (please complete the following information): Linux skydancer 5.11.0-49-generic #55-Ubuntu SMP Wed Jan 12 17:36:34 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux

Additional context

• cli tigris version v1.0.0-alpha.8
• tigris container date docker inspect --format '{{ .Created }}' tigrisdata/tigris:latest 2022-05-04T01:53:00.992143876Z
• github.com/tigrisdata/tigris-client-go v1.0.0-alpha.10

Describe the bug Containers should have resource limits

deployment/manifests/base/deployment.yaml deployment/manifests/base/search-deployment.yaml

Expected behavior Without resource limits, containers can starve other processes

Allowing filtering and sorting by metadata fields created_at and updated_at:

{
    "q": "",
    "filter": {
        "$and": [
            {
                "created_at": {
                    "$gt": "2022-08-16T17:29:28.000Z"
                }
            },
            {
                "created_at": {
                    "$lt": "2022-08-25T17:29:28.000Z"
                }
            }
        ]
    }
}

TODO:

• [x] setup context for stream server interceptor
• [ ] make the cache reload when namespace is created
• [x] add tests.
• [x] check if there is anything needs to be done for forwarders
• [x] fix tests
• [x] fix lint warnings

https://github.com/tigrisdata/tigris/blob/main/scripts/install_build_deps.sh seems to not support windows.

is that because foundation db just does not run on windows ?

Is your feature request related to a problem? Please describe. This feature would allow a user to store data as a BLOB type, or possibly as a pointer to an S3 object store, for example.

Describe the solution you'd like Example: need to store image and video files (easily 100KB+ to 1GB+ in size). Tigris and FDB are great to store metadata in (and often the metadata is just as important as anything) but it would be highly desirable to store blobs (in PostgreSQL bytea type, for example) as a binary large object. Alternative, it could interesting to integrate something with, for example, MinIO or S3 where Tigris has the metadata and index but the blobs are stored into an S3 location.

Describe alternatives you've considered MinIO, large table support such as HBase/Accumulo or Apache Iceburg with HDFS or S3 for file storage

Additional context n/s

Is your feature request related to a problem? Please describe. It's a feature request to let know the collection size

Describe the solution you'd like

• Approximate or precise size of the collection in the in a new method totalSize in a /api/v1/databases/{db}/collections/{collection}/ namespace
• Includes all: indexes, keys, fields, etc.
• Implementations in the tigrisdb and support it in the client

Describe alternatives you've considered

• Just to start a 0 value can be returned. The actual size value can be added later.
• Give a size value in the ListCollectionsResponse method is not really meant for presenting the size information

Additional context