Skip to content

Add I/O timeout handling and reporting for hung (stuck) I/Os #2012

@aqibahmad

Description

@aqibahmad

Currently, we have a max_latency parameter that only works for completed I/Os and generates error if a completed I/O operation exceeded the time specified by this parameter. Therefore, it is kind of a pass/fail criteria on completed I/O requests.

Consider a situation where FIO is not able to complete an I/O operation within a particular amount of time. This could be because the underlying filesystem (local or NFS) is hung and not responding. In this case, it would be really beneficial to generate errors with FIO. For example: a parameter called stuck_timeout=30sec. This implies that if no response is received within 30 seconds, FIO will generate an I/O error with details of the operation that failed.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementpatcheswelcomeNo one internal is working on this but an external contributor's help is very welcome!

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions