[LUDOC-50] The Lustre Ops Manual should doc LNET/LND options for peer_timeout and avoid_asym_router_failure

• Type: Improvement
• Resolution: Fixed
• Priority: Minor
• Labels: LNET QContent
• Business Value: 14
• Assignee: Linda Bebernes (Inactive)
• Reporter: Cory Spitz

Description

Peer health detection and asymmetric router failure detection avoidance options are not documented in the Lustre Ops Manual.

Bugzilla 24404 (https://bugzilla.lustre.org/show_bug.cgi?id=24404) provides some guidance for these options and also calls out some other undocumented LNET/LND option that need improvement.

Comments

[Linda Bebernes (Inactive)]

Changes reviewed and merged at http://review.whamcloud.com/#/c/8127/. Resolved.

[Linda Bebernes (Inactive)]

This ticket covers several topics -

• LNET peer health (peer_timeout, avoid_asym_router_failure - a draft of this content has been created and is being reviewed by Amir.

• LNET router buffers (tiny, small, large) - this has been pushed to gerrit for review at http://review.whamcloud.com/#/c/7997/

• LNET message credits - this will be covered in LUDOC-189 - Document LNET Message Credits

[Cory Spitz]

The module info shows: "Avoid asymmetrical failures: reserved, use at your own risk". I'm not sure if that needs to be the case, but I believe that avoid_asym_router_failure should be documented nonetheless.