disttask: add local disk space to node resource managed by DXF #59607
Conversation
ti-chi-bot
bot
added
release-note-none
|
Hi @tangenta. Thanks for your PR. PRs from untrusted users cannot be marked as trusted with I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #59607 +/- ##
================================================
+ Coverage 73.0112% 74.6811% +1.6698%
================================================
Files 1694 1742 +48
Lines 468361 479073 +10712
================================================
+ Hits 341956 357777 +15821
+ Misses 105382 98726 -6656
- Partials 21023 22570 +1547
Flags with carried forward coverage won't be shown. Click here to find out more.
|
| } | ||
|
|
||
| // GetTaskDiskResource gets available disk for a task. | ||
| func (nr *NodeResource) GetTaskDiskResource(concurrency int, quotaHint uint64) uint64 { |
There was a problem hiding this comment.
is quotaHint used to allocate less resource for some task? why do we need it if we cannot allow some task to alloc more resource?
and it should be part of StepResource, no need a special API
There was a problem hiding this comment.
StepResource is only used by the step executor. We need to know the resource in the scheduler before splitting subtasks.
There was a problem hiding this comment.
is quotaHint used to allocate less resource for some task?
Yes.
why do we need it if we cannot allow some task to alloc more resource?
Because other factors may affect the resource allocation, for example, @@tidb_ddl_disk_quota can change dynamically.
There was a problem hiding this comment.
if the resource per slot for a subtask is always <= the mount of disk per slot = D, seems ok to just use D, no need to adjust
There was a problem hiding this comment.
We can't determine which one is larger, tidb_ddl_disk_quota or actual disk capacity. For example, a user has a 2TB node, but only wants to use 500GB for DXF, then the user can set tidb_ddl_disk_quota to 500GB
There was a problem hiding this comment.
no need to consider tidb_ddl_disk_quota, we can always use the total_disk we got by inspecting disk stat.
| nodeRes := proto.NewNodeResource(16, 32*units.GiB) | ||
| exeMgr, err := taskexecutor.NewManager(c.Ctx, id, c.TaskMgr, nodeRes) |
There was a problem hiding this comment.
in TestDXFContext, the cpu count is a input param that's passed to github.com/pingcap/tidb/pkg/util/cpu/mockNumCpu",, the new node need also conform to this
see newTestDXFContext
ti-chi-bot
bot
added
approved
needs-1-more-lgtm
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: CbcWestwolf, D3Hunter The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
ti-chi-bot
bot
added
lgtm
and removed
needs-1-more-lgtm
[LGTM Timeline notifier]Timeline:
|
What problem does this PR solve?
Issue Number: ref #49008
Problem Summary:
We didn't consider disk space for each subtask when implementing DXF. In the following enhancement, we may utilize the local storage for global sort, it requires managing the disk usage resource.
What changed and how does it work?
NodeResource.NodeResourcefrom packagetaskexecutortoproto.Check List
Tests
Side effects
Documentation
Release note
Please refer to Release Notes Language Style Guide to write a quality release note.