Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

DAOS-9883 downgrade libfabric to 1.12 and varia #16

Merged
merged 3 commits into from
Feb 10, 2022
Merged

Conversation

johannlombardi
Copy link

Downgrade libfabric from 1.14 to 1.12 to address performance issue.
Adjust a few yaml parameter for 2.0
Set first core to 0.
Use single thread dfuse for io500.
Bump default aggregation threshold to meet DAOS 1.2 values.

Signed-off-by: Johann Lombardi [email protected]

Downgrade libfabric from 1.14 to 1.12 to address performance issue.
Adjust a few yaml parameter for 2.0
Set first core to 0.
Use single thread dfuse for io500.
Bump default aggregation threshold to meet DAOS 1.2 values.

Signed-off-by: Johann Lombardi <[email protected]>
Comment on lines 83 to 84
rpm -i --force ./libfabric-1.12.0-1.el7.x86_64.rpm
rpm --erase --nodeps libfabric-1.14.0

Choose a reason for hiding this comment

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

FWIW, there is yum downgrade to downgrade packages without the potential of breaking things the way rpm --force could.

@@ -123,7 +123,7 @@ daos cont get-prop ${POOL_LABEL} ${CONT_LABEL}
log "Use dfuse to mount ${CONT_LABEL} on ${IO500_RESULTS_DFUSE_DIR}"
pdsh -w ^hosts sudo rm -rf "${IO500_RESULTS_DFUSE_DIR}"
pdsh -w ^hosts mkdir -p "${IO500_RESULTS_DFUSE_DIR}"
pdsh -w ^hosts dfuse --pool="${POOL_LABEL}" --container="${CONT_LABEL}" --mountpoint="${IO500_RESULTS_DFUSE_DIR}"
pdsh -w ^hosts dfuse -S --pool="${POOL_LABEL}" --container="${CONT_LABEL}" --mountpoint="${IO500_RESULTS_DFUSE_DIR}"
Copy link
Contributor

Choose a reason for hiding this comment

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

@johannlombardi Does this change the perf per client in GCP?

Copy link
Author

Choose a reason for hiding this comment

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

Not for io500 since it does not use dfuse. For app using dfuse directly, -S should not be used.

Kmannth
Kmannth previously approved these changes Feb 9, 2022
Copy link
Contributor

@Kmannth Kmannth left a comment

Choose a reason for hiding this comment

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

Looks good to me.

cboneti
cboneti previously approved these changes Feb 10, 2022
@markaolson markaolson dismissed stale reviews from cboneti and Kmannth via 0fb1fce February 10, 2022 19:08
@johannlombardi johannlombardi merged commit 21b1f49 into master Feb 10, 2022
@johannlombardi johannlombardi deleted the 20_update branch February 10, 2022 19:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Development

Successfully merging this pull request may close these issues.

5 participants