From 47c2ebb6569a4aac60c595942a7df11b50c25906 Mon Sep 17 00:00:00 2001 From: ehb54 Date: Mon, 27 Jul 2026 14:59:27 -0500 Subject: [PATCH] elasticmanager: disable the cpu threshold alert, it cannot work The first real between frame gap produced a false positive. Slot 2 was warned about at 18:08 and recovered on its own at 19:39, having been perfectly healthy the whole time. The gap with no waxsis container ran 1h41m; the alert fired 25 minutes in and stood for 1h31m. The threshold was sized expecting gaps around ten minutes. This one was ten times that. More importantly it is not a tuning problem. During the gap slot 2 read 2%. Genuinely idle slot 1 reads 2% on every sample in the file. A healthy job between frames and an abandoned instance are numerically identical from the instance, because while the job does its local work in the saxsafold container the instance really is doing nothing, which is also what a stranded instance is doing. There is no signal there to threshold, and raising consecutive past this gap only trades a false positive for a longer blind window, with no safe value since gap length is set by work happening elsewhere. Removes probe:min_pct so nothing is raised. Sampling continues and em_probe.log keeps accumulating, which is the second time that data has corrected a guess. probe_check() stays, inert without min_pct, since its counter and one-alert-per-episode handling are reusable once the trigger is a signal that actually separates the two cases: the acquiring pid, carried in the tag, which is a fact rather than an inference. Co-Authored-By: Claude Opus 5 --- languages/html5/openstack/elasticmanager/em_config.json | 2 -- 1 file changed, 2 deletions(-) diff --git a/languages/html5/openstack/elasticmanager/em_config.json b/languages/html5/openstack/elasticmanager/em_config.json index 4f0ac77a..497142ed 100644 --- a/languages/html5/openstack/elasticmanager/em_config.json +++ b/languages/html5/openstack/elasticmanager/em_config.json @@ -17,8 +17,6 @@ ,"probe" : { "interval" : 300 ,"history" : "em_probe.log" - ,"min_pct" : 50 - ,"consecutive" : 3 } ,"flavors" : { "m3.2xl" : {