milvus | native vector database , storage for next generation AI | Search Engine library

I follow this link to install milvus cluster. When I execute this command "helm install my-release milvus/milvus" ,I find that a pod never started successfully. enter image description here

my-release-etcd-0 1/1 Running 0 3m58s

my-release-milvus-standalone-8587d4796d-r579n 0/1 Running 0 3m58s

my-release-minio-54fc79dbdf-gzlsh 1/1 Running 0 3m58s

[2022/01/21 07:30:00.210 +00:00] [DEBUG] [client.go:82] ["DataCoordClient, not existed in msess "] [key=datacoord] ["len of msess"=0] [2022/01/21 07:30:00.209 +00:00] [ERROR] [client.go:115] ["failed to get client address"] [error="number of datacoord is incorrect, 0"] [stack="github.com/milvus-io/milvus/internal/util/grpcclient.(*ClientBase).connect\n\t/go/src/github.com/milvus-io/milvus/internal/util/grpcclient/client.go:115\ngithub.com/milvus-io/milvus/internal/util/grpcclient.(*ClientBase).GetGrpcClient\n\t/go/src/github.com/milvus-io/milvus/internal/util/grpcclient/client.go:87\ngithub.com/milvus-io/milvus/internal/util/grpcclient.(*ClientBase).callOnce\n\t/go/src/github.com/milvus-io/milvus/internal/util/grpcclient/client.go:177\ngithub.com/milvus-io/milvus/internal/util/grpcclient.(*ClientBase).ReCall\n\t/go/src/github.com/milvus-io/milvus/internal/util/grpcclient/client.go:217\ngithub.com/milvus-io/milvus/internal/distributed/datacoord/client.(*Client).GetComponentStates\n\t/go/src/github.com/milvus-io/milvus/internal/distributed/datacoord/client/client.go:110\ngithub.com/milvus-io/milvus/internal/util/funcutil.WaitForComponentStates.func1\n\t/go/src/github.com/milvus-io/milvus/internal/util/funcutil/func.go:50\ngithub.com/milvus-io/milvus/internal/util/retry.Do\n\t/go/src/github.com/milvus-io/milvus/internal/util/retry/retry.go:34\ngithub.com/milvus-io/milvus/internal/util/funcutil.WaitForComponentStates\n\t/go/src/github.com/milvus-io/milvus/internal/util/funcutil/func.go:74\ngithub.com/milvus-io/milvus/internal/util/funcutil.WaitForComponentHealthy\n\t/go/src/github.com/milvus-io/milvus/internal/util/funcutil/func.go:89\ngithub.com/milvus-io/milvus/internal/distributed/querycoord.(*Server).init\n\t/go/src/github.com/milvus-io/milvus/internal/distributed/querycoord/service.go:183\ngithub.com/milvus-io/milvus/internal/distributed/querycoord.(*Server).Run\n\t/go/src/github.com/milvus-io/milvus/internal/distributed/querycoord/service.go:95\ngithub.com/milvus-io/milvus/cmd/components.(*QueryCoord).Run\n\t/go/src/github.com/milvus-io/milvus/cmd/components/query_coord.go:50\ngithub.com/milvus-io/milvus/cmd/roles.(*MilvusRoles).runQueryCoord.func1\n\t/go/src/github.com/milvus-io/milvus/cmd/roles/roles.go:178"]

I know that when dockers installs Milvus, port 19530 is the Milvus port. I would like to know what port 19121 does and if it can be installed without mapping this port？ The dockers command to install milvus is as follows：

I have a question about distance calculation in Milvus. In Milvus, I use the L2 distance calculation to query a vector for top1 and Milvus returns a distance of 9.340524. whereas the distance I get between the query vector and the return using the L2 formula is 2.156227. Why is the formula for distance calculated differently from the result returned by Milvus?

1. When calling the drop_collection() to delete a table, why is memory not released?

2. How could milvus free up the memory space occupied by the deleted collection?

I installed it by following the instruction from Milvus. But every time I installed it with docker, it failed to start.

The docker command is as followed:

I have a macro which runs through several pivot tables. In this one if the option isn't available e.g. I'm trying to select the number 50, but that isn't in the pivot table data. I want it to select blank instead. I've done an On error GoTo This works when the code is not there, but if the code IS there it acts like it isn't and returns blank values in the filter. Can anyone tell what I've done wrong here?

I'm using milvus to make image similarity research in a dataset of around one million images.

The basic setting is :

• milvus in one docker
• mysql in an other docker
• milvus reads/writes in the mysql database.

Question : is it reasonable to have mysql running on a remote machine ? How many mysql requests does Milvus do for performing one research ?

If not, I guess that the most reasonable path is to copy the mysql docker with its data from the "train machine" to the "search machine". Correct ?

This question is related to this one.

I'm using milvus to make image similarity research in a dataset of around one million images.

When I insert an image, Milvus returns an id which is the id of that image in the milvus database (mysql). When Milvus performs a research, it returns the id of the most similar image.

Now I have to keep track of the relation "id -> image name" to be able to present the correct image when the research is done.

What is done in the furnished examples is to store that relation in a local dictionary (or using a package like diskcache).

This works well when the training is done on the same machine as the research.

I want to train on one machine and make the inference on an other one.

My first idea would be to store the relation "id->name" in a separate table on the same mysql as the rest of milvus.

Is that a good idea ? That would cost one more mysql request after the research.

Can I make a mysql join between the milvus's table "id->vector" and my table "name->id", so that milvus returns its image id in the same time as my image name ?