Measure each item its own way; never compare a partial size (hq ADR 0233)
A walk over a large library every hour loads the array that protects it. An item now says how it is measured — a bounded daily walk, a dataset's counters, or its top level only — and a size that is a lower bound is kept as such and never read as a shrink.
This commit is contained in:
@@ -38,6 +38,8 @@ create table data_item (
|
||||
last_backup timestamptz,
|
||||
-- What the holder could not measure, when it could not: the path gone, a walk refused.
|
||||
measure_error text,
|
||||
-- What the newest size is: exact, a dataset's whole size, partial (a lower bound) or none.
|
||||
precision text,
|
||||
-- The redundant storage it is on, as the holder read it: zfs, md or btrfs, which pool or device,
|
||||
-- whether it is healthy, and what it said.
|
||||
redundancy_kind text,
|
||||
@@ -53,7 +55,8 @@ create table data_item (
|
||||
);
|
||||
|
||||
-- One row per measurement kept, at most one an hour per item, for ninety days: what a shrink is
|
||||
-- read against.
|
||||
-- read against. Only a comparable size is kept: exact, or a dataset's own counters — never a partial
|
||||
-- walk's lower bound.
|
||||
create table data_reading (
|
||||
machine text not null,
|
||||
module text not null,
|
||||
|
||||
Reference in New Issue
Block a user