close: #3416

Fixes: #3416

Motivation

I use the mdtest with CubeFS backend support to measure the throughput of CubeFS meta subsystem. Each mdtest process will create its own CubeFS client with the MetaWrapper having the same epoch.

Each mdtest process creates the same directory tree (one dir contains millions of files). In this situation, all mdtest processes will create their root dir inode in the same meta partition, because all the mdtest processes' CubeFS client have the same epoch. Then all the files' dentry will be created in the same meta partition, which will cause significant performance degradation.

Modifications

In the NewMetaWrapper function, I assign a random value to the mw.epoch. The value is in the range of [0, len(mw.rwPartitions)].

Types of changes

• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Bugfix (non-breaking change which fixes an issue)
• Documentation Update (if none of the other choices apply)
• So on...

Verifying this change

• Make sure that the change passes the testing checks.

This change is a trivial rework / code cleanup without any test coverage.

(or)

This change is already covered by existing tests, such as (please describe tests).

(or)

This change added tests and can be verified as follows:

(example:)

• This can be verified in development debugging
• This can be realized in a mocked environment, like a test cluster consisting in docker

(or)

This change MUST reappear in online clusters, or occur in that specific scenarios.

Does this pull request potentially affect one of the following parts:

• Master
• MetaNode
• DataNode
• ObjectNode
• AuthNode
• LcNode
• Blobstore
• Client
• Cli
• SDK
• Other Tools
• Common Packages
• Dependencies
• Anything that affects deployment

Documentation

• doc
• doc-required
• doc-not-needed
• doc-complete

Review Expection

• in-two-days
• weekly
• free-time
• whenever