Skip to content

kubemark cluster starts partition servers in parallel#1113

Merged
centaurus-cloud-bot merged 4 commits intoCentaurusInfra:masterfrom
h-w-chen:hw-kubemark-start-xp-in-parallel
Jul 13, 2021
Merged

kubemark cluster starts partition servers in parallel#1113
centaurus-cloud-bot merged 4 commits intoCentaurusInfra:masterfrom
h-w-chen:hw-kubemark-start-xp-in-parallel

Conversation

@h-w-chen
Copy link
Collaborator

@h-w-chen h-w-chen commented Jul 6, 2021

What type of PR is this?
/kind feature

What this PR does / why we need it:
This PR adopts parallel starting of arktos scale-out partitions when starting the kubemark cluster.

Current start-kunemark scripts starts the scale-out partitions (TP or RP) in sequence, which is quite time consuming if the number of TP servers and that of RP servers are not minimum. There are various ways to optimize the cluster setup time and reduce the overhead; one obvious way is making these partition servers started in parallel can significantly reduced the cluster setup time. This PR starts multiple TPs in parallel, after all TPs are started, then starts multiple RPs in parallel.

This PR does not try to start TPs and RPs at the same time.

Does this PR introduce a user-facing change?:
NONE

@h-w-chen h-w-chen requested review from Sindica, q131172019 and yb01 July 6, 2021 20:56
@h-w-chen h-w-chen changed the title Hw kubemark start xp in parallel kubemark cluster starts partition servers in parallel Jul 6, 2021
#
MASTER_METADATA=""

### to upload etcd image / binary tar once
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

the test_resource_upload() can be added to the very beginning of the test setup and ensure it is done. so that this check can be removed

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

admin cluster also needs this uploading, and it is directly set up by kube-up.sh script; starting partition clusters also eventually call into kube-up.sh; if we don't want to change this structure, we have to have some sort of condition check in place.

Copy link
Collaborator

@yb01 yb01 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

/lgtm
with a few minor comments

@zmn223
Copy link
Collaborator

zmn223 commented Jul 13, 2021

/lgtm

@zmn223
Copy link
Collaborator

zmn223 commented Jul 13, 2021

/lgtm

@zmn223
Copy link
Collaborator

zmn223 commented Jul 13, 2021

/approve

@centaurus-cloud-bot
Copy link
Collaborator

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: yb01, zmn223

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details Needs approval from an approver in each of these files:

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

@centaurus-cloud-bot centaurus-cloud-bot merged commit 1dfd6c4 into CentaurusInfra:master Jul 13, 2021
@h-w-chen h-w-chen deleted the hw-kubemark-start-xp-in-parallel branch July 13, 2021 19:58
Sindica added a commit that referenced this pull request Jul 17, 2021
* design doc: DaemonSet support in scale-out Arktos (#1109)

* doc: DaemonSet support in scale-out Arktos

* minor: rephrased daemonset managability of failed TP

* added notes of scale-up arktos supporting system tenanted DS

* put daemonset design doc in separate file

* trivial: resource manager reworded as resource partition

* added design alternatives based on peer feedback

* emphasized on DS and supporting resources in unit of TP; put more detail of kubelet secret/configmap local store changes

* added clarifications based on peer feedback

* minor: revised based on peer review

* kubemark cluster starts partition servers in parallel (#1113)

* multiple partitions of same kind (tp/rp) able to start in parallel

* eliminates /tmp/saved_tenant_ips.txt and TP_IP_CONCAT var from kubemark setup scripts

* minor: todo comments for dedicated log stream of parallel calls

* Bump Arktos to v0.8.0 (#1116)

* Fix a bug that event client was created with wrong user agent (#1120)

Co-authored-by: hwchen <hong.chen@futurewei.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants