10

For the testing of a new traffic management tool there was an entry on the recently renewed stackstatus.net. I remember clearly because I asked about that posting in my answer here.

Looking back today the entry is no longer visible at the home-page, which is fine as the first step of the test was completed. However there is also no entry of this in the incident history.

What is the current guideline in place to deal with maintenance notifications that have "expired"?

5
  • "which is fine as the first step of the test was completed" - it wasn't there even before the test was started, I checked the site daily and never saw it, actually. Jun 20, 2022 at 12:33
  • I knew it was there, as I asked about it in my answer on that announcement @ShadowWizardSaysNoMoreWar
    – Luuklag
    Jun 20, 2022 at 12:35
  • Yeah, I don't claim it was never there, just that it was removed on the same day it was posted, as far as I can tell. :) Jun 20, 2022 at 12:37
  • 1
    I too remember it was there. I saw the link (in that announcement) in plain text and was going to bring it up, but you already had..
    – Justin
    Jun 20, 2022 at 13:57
  • 2
    Interestingly enough, the entry is still accessible by direct URL (my RSS feeder still has it!), even though it doesn't exist in the history list.
    – zcoop98
    Jun 20, 2022 at 22:39

1 Answer 1

5

To address "why don't I see an event from June on the home page?"

The incident history page is configured to show a quarter at a time. So if you click this left arrow at the top:
enter image description here

You'll end up here: https://www.stackstatus.net/history?start=2022-04-01&end=2022-06-30

Which does show this test from June:

enter image description here

Arguably, the UX is not the most intuitive, nor consistent with our own UX, but we'll need to work with the vendor over time to make usability improvements and other styling changes.

To address "why is the June 2nd event showing and not the June 16th event?"

  • The June 16th test didn't have the right severity, so publishing to the status page was skipped.
  • The June 2nd event also didn't have the right severity, but it was published to the status page while the runbook was still being configured. Conditions that should have prevented it from being published didn't exist yet.

Working on this now with the vendor to correct both events.

13
  • Aaron, that is correct. But at the time I posted this question, June 20th, we should've arguably been in that quarter, and it should've been visible then. I didn't take a screenshot then, so obviously at some point it started showing up in the incident history.
    – Luuklag
    Sep 22, 2022 at 12:35
  • 1
    I don't know what the page looked like then either, but probably things were still being wired up and glued together.
    – Aaron Bertrand Staff
    Sep 22, 2022 at 12:36
  • 1
    @AaronBertrand I remember, it showed one day, and you could go back and forth without any limit, I viewed incidents in 2050 for example. And in 1600. Luckily those were incident free days. Anyhow, I'm 100% sure I viewed June 2nd (2022 ;)), and it was empty. The bug reported here isn't about display, it's about the missing content. Sep 22, 2022 at 12:41
  • 1
    @ShadowTheKidWizard Right, I just don't know how we can verify a missing content bug from June, I posted an answer because regardless of what kind of display issues were happening in June, the problem seems to be fixed now. And also because multiple comments here and elsewhere imply that the navigational element to move back to previous quarters was not prominent/obvious.
    – Aaron Bertrand Staff
    Sep 22, 2022 at 12:42
  • Maybe this will shed some light? Date is different, showing 16th. Which is the correct date, by the way. The current state is wrong, it wasn't done on June 2nd. Sep 22, 2022 at 12:43
  • @ShadowTheKidWizard I don't think that's "wrong", I think it's a different entry altogether. Totally agree that it's weird though, and this answer does not address that missing entry at all.
    – zcoop98
    Sep 22, 2022 at 16:20
  • 2
    @zcoop98 Still investigating. There were two "test" incidents in June, the wrong one is showing in the history, I'm working with the vendor to correct them both.
    – Aaron Bertrand Staff
    Sep 22, 2022 at 16:35
  • 1
    There seems to be a lot of anger about the accuracy of a test event that appeared on a status page three months ago while the status page was first being configured.
    – Aaron Bertrand Staff
    Sep 22, 2022 at 16:43
  • 1
    Thanks for looking into all this @AaronBertrand, I know it can be a pain working with external vendors for these kind of things. So Kudo's to you for diving into it.
    – Luuklag
    Sep 22, 2022 at 18:45
  • @AaronBertrand Sweet, appreciate the update! Is it normal then to expect more entries in the future that won't publish to the list? The only reason I came by that "missing" entry in the first place was through my RSS feed reader that caught it.
    – zcoop98
    Sep 22, 2022 at 22:32
  • 1
    @zcoop98 It’s our hope that for all future planned maintenance that get a notification here, those incidents will also make the status page (and hence added to the history list) automatically. It just may take a few incidents before we perfect the process. I wasn’t involved with the June 2/16 incidents - when I first started looking earlier today I was convinced that Shadow was right, that they were the same incident, but after digging deeper I found they were separate. That one made the list and the other didn’t is just an indicator of early process tuning. We’ll fix them and “get there.”
    – Aaron Bertrand Staff
    Sep 22, 2022 at 22:38
  • 1
    I know how to remove the June 2nd event from the list, and I plan to do so, because it shouldn’t be there - it was literally just a test of pushing something to the page. The June 16th test was the one Josh announced here and had more visibility. Being totally honest here: I don’t want to remove the June 2nd event until I figure out how to add the June 16th event. I’ve asked for a simple toggle control that would let us manually ignore the workflow / conditional runbook logic that evaluates whether an incident makes the page, but that was just today, so…
    – Aaron Bertrand Staff
    Sep 22, 2022 at 22:42
  • 1
    A little more info: There are two ways to created a planned maintenance event in the system. One is to start a scheduled maintenance window, and the other is to create a P-level incident and tag it with maintenance. We used the latter (because it’s the only one with a runbook attached), and that is the one the vendor didn’t expect us to use for this purpose, so it is a little more cumbersome to go back three months and reverse engineer how it got into this state (did the runbook change? did the tags or severity change? Etc).
    – Aaron Bertrand Staff
    Sep 22, 2022 at 22:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .