Skip to content

Conversation

@rhc54
Copy link
Contributor

@rhc54 rhc54 commented May 30, 2019

The debrujin component is using an algorithm that doesn't respect the
previously assigned parent ID. This causes the other components to have
their routing trees broken whenever debrujin updates routes. This
happens whenever more than 256 nodes are involved, thus breaking tree
spawn for sizes >= 256

Thanks to @zrss for the report and diagnosis!

Signed-off-by: Ralph Castain [email protected]

The debrujin component is using an algorithm that doesn't respect the
previously assigned parent ID. This causes the other components to have
their routing trees broken whenever debrujin updates routes. This
happens whenever more than 256 nodes are involved, thus breaking tree
spawn for sizes >= 256

Thanks to @zrss for the report and diagnosis!

Signed-off-by: Ralph Castain <[email protected]>
@rhc54 rhc54 added the bug label May 30, 2019
@rhc54 rhc54 added this to the v3.0.5 milestone May 30, 2019
@rhc54 rhc54 requested a review from jsquyres May 30, 2019 13:15
@rhc54 rhc54 self-assigned this May 30, 2019
@bwbarrett
Copy link
Member

We talked about this change (and related PRs on the v3.1.x and v4.0.x branches) on the call today and decided that removing the component is the right path forward. While we try very hard to avoid removing components in the middle of a release series, in this case we are going to do so. The component never worked properly in this release series, because of the stateful component selection and the bug in component selection logic for this component, and because of the cost of a proper fix, this is the right choice.

@bwbarrett bwbarrett merged commit 77567ad into open-mpi:v3.0.x Jun 4, 2019
@rhc54 rhc54 deleted the cmr30/routed branch November 27, 2019 19:17
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.

2 participants