Struct Faults 

pub struct Faults {

    pub fail_next: usize,
    pub refuse_file_uploads: usize,
    pub panic_file_uploads: usize,
    pub read_panics: BTreeSet<ObjectPath>,
    pub read_failures: BTreeMap<ObjectPath, StorageFailure>,
    pub status_failures: BTreeMap<ObjectPath, StorageFailure>,
    pub delete_failures: BTreeMap<ObjectPath, StorageFailure>,
    pub create_failures: BTreeMap<ObjectPath, (StorageFailure, u16)>,
    pub list_failures: BTreeMap<ObjectPrefix, StorageFailure>,
    pub read_chunk_size: usize,
    pub body_failures: BTreeMap<ObjectPath, BodyFailure>,
    pub garbled_reads: BTreeMap<ObjectPath, Garble>,
    pub failure: StorageFailure,
    pub response: Option<u16>,
    pub retry_after: Option<u64>,
    pub delay: Duration,
    pub drop_part: bool,
    pub lose_part_reply: bool,
    pub lose_initiation_reply: bool,
    pub lose_completion_reply: bool,
    pub abort_failures: usize,
    pub expire_uploads: bool,
    pub overwrite_create: bool,
    pub hide_listings: bool,
    pub listing_lag: Option<Duration>,
    pub fail_duplicate_cleanup: bool,
    pub refuse_after_storing: usize,
    pub refuse_replacements: usize,
    pub grant_failure: Option<StorageFailure>,
}

Faults applied to the next calls, including accepted parts with lost replies.

Fields

fail_next: usize

Fail this many calls before doing their work.

refuse_file_uploads: usize

Refuse this many file uploads as they start, each with failure, response and retry_after as fail_next fails a call: a file’s single create, or the call beginning its multipart upload. Nothing is stored. Every other call goes through, so a pass beside the upload can’t take the fault.

panic_file_uploads: usize

Panic in this many file uploads as they start, as a bug in the caller’s upload would; every other call goes through.

read_panics: BTreeSet<ObjectPath>

Panic in exact reads at these paths, as a bug in the reader would.

read_failures: BTreeMap<ObjectPath, StorageFailure>

Fail exact reads at these paths, retaining other in-flight responses.

status_failures: BTreeMap<ObjectPath, StorageFailure>

Fail exact status calls at these paths, independently of reads.

delete_failures: BTreeMap<ObjectPath, StorageFailure>

Fail exact deletions at these paths; the object stays.

create_failures: BTreeMap<ObjectPath, (StorageFailure, u16)>

Refuse creates at these paths with a provider response of this HTTP status; nothing is stored.

list_failures: BTreeMap<ObjectPrefix, StorageFailure>

Fail listings of exactly these prefixes.

read_chunk_size: usize

Maximum bytes returned by each body read; nonzero.

body_failures: BTreeMap<ObjectPath, BodyFailure>

Failures after metadata, independently of request-start failures.

garbled_reads: BTreeMap<ObjectPath, Garble>

Change one byte of responses at these paths, each response differently, as faults on the way to the device would; the stored object stays as it was.

failure: StorageFailure

The classification of injected failures.

response: Option<u16>

Injected failures arrive as a provider response with this HTTP status (E5). None, without retry_after, is a request that received no response.

retry_after: Option<u64>

Injected failures arrive as a response, a 429 unless response names another status, asking every consumer to wait this many seconds (Retry-After).

delay: Duration

Delay each call without blocking its executor.

drop_part: bool

Drop the next part before it is stored.

lose_part_reply: bool

Store the next part, but lose the reply before advancing the caller’s session.

lose_initiation_reply: bool

Start the next planned session’s upload, but lose the reply: the caller never learns the id of the session the provider holds.

lose_completion_reply: bool

Publish the next upload but lose the completion reply.

abort_failures: usize

Refuse this many upload aborts, retaining the pending bytes.

expire_uploads: bool

Expire all pending sessions before the next call.

overwrite_create: bool

Accept creates at occupied paths, violating the provider contract.

hide_listings: bool

Hide published objects from listings while exact reads and status still see them.

listing_lag: Option<Duration>

Hide objects no older than this lag; exact reads still see publication.

fail_duplicate_cleanup: bool

Refuse a duplicate create but fail to abort its upload session.

refuse_after_storing: usize

Store this many creates and replacements, then answer each with a 429, as a copy publication does when its temporary’s deletion is refused after the object is published (§20.6).

refuse_replacements: usize

Refuse this many conditional replacements with Conflict although they name the current revision, leaving the object unchanged, as OneDrive can refuse every writer of one race.

grant_failure: Option<StorageFailure>

Fail every grant of access to an account with this failure, as a provider refusing to share the store would.

Implementations

impl Faults

pub fn none() -> Self

No failures or delays.

Trait Implementations

impl Clone for Faults

fn clone(&self) -> Faults

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.