jschoubben
53a79a17a1
The host's error text may carry a duration or a counter, and a resource looping on it would never have read as stuck. The previous row is read and compared here. Stuck needs a start to say. A container may mount the file a binding lands in; the runtime socket is declared under both of its spellings; the catalogue-wide test takes MESH_CATALOG.
17 lines
1.1 KiB
SQL
17 lines
1.1 KiB
SQL
-- A failure that repeats is told apart from one that just happened (novox/hq 04-ISSUES/065).
|
|
--
|
|
-- A node re-applies what it holds on a steady interval and reports each time (novox/hq ADR 0010),
|
|
-- which is right for a failure that goes away by itself -- the overlay not up yet, a registry
|
|
-- briefly unreachable -- and makes a failure that will never go away look exactly the same: one
|
|
-- row, replaced, saying "failed" at a fresh time. Nothing distinguished "failed once, will succeed
|
|
-- when its dependency arrives" from "failed identically for ever", and nothing escalated the second.
|
|
--
|
|
-- Still one row per node. What is added is how long the CURRENT failure has been the same one:
|
|
-- when it first appeared, and how many reports in a row have said it -- the same outcome, the same
|
|
-- refusal, the same failed resources by id (not by the host's words, which may carry a duration).
|
|
-- A report that says something different starts the count again; a clean apply clears it.
|
|
|
|
alter table node_report
|
|
add column failing_since timestamptz,
|
|
add column failures int not null default 0;
|