<DasSkelett[m]>
Okay I think I tested now pretty much everything, mod creation, update, autoupdate, download, publishing, locking, unlocking, following, unfollowing, getting email notifications, even account creation.
<DasSkelett[m]>
Hope there's no bug left hiding in a corner.
<VITAS[m]>
<is building new errorpages for the proxy that wont terrify you as much with "all is borken technical alarm!"
DasSkelett[m] has quit [Quit: Client limit exceeded: 6]
<HebaruSan[m]>
Honestly the best place to announce upcoming downtimes is a header on spacedock.info itself
<HebaruSan[m]>
Maybe that could be integrated into the blogging function
<HebaruSan[m]>
Go to admin page, start a blog post, type description of downtime, check a new checkbox "make this a global announcement", submit
<VITAS[m]>
yes but blog as well as sd can be offline while the announcment applies
<HebaruSan[m]>
I'm thinking about announcing beforehand
<HebaruSan[m]>
So people would know to expect it
<VITAS[m]>
yes im with you on the solution for announcing it
<HebaruSan[m]>
I thought you were working on a new 404 page for during the downtime
<VITAS[m]>
i did
<VITAS[m]>
its what apache webserver and ats show when sd is down
<HebaruSan[m]>
Then after the downtime, go back into that blog post, edit, uncheck the checkbox, save
<VITAS[m]>
i want to tell them to check page x for announcments
<HebaruSan[m]>
Nobody will do that tho
<VITAS[m]>
if they misse dit and whant to know whne it sover and so on
<VITAS[m]>
it sbetter than "this site isnt there"
<HebaruSan[m]>
That would be like fifteen words to add to the downtime page
<HebaruSan[m]>
Why make people click?
<VITAS[m]>
because i would have to edit 30 pages
<VITAS[m]>
in 3 places
<VITAS[m]>
each time
<HebaruSan[m]>
Are you planning to use the same page for all of your other hosted sites then?
<VITAS[m]>
no
<HebaruSan[m]>
What are the 30 pages then? I don't quite follow that
<VITAS[m]>
but there are different kinds of errors
<VITAS[m]>
apache webserver != ats
<VITAS[m]>
and so on
<HebaruSan[m]>
Now I am more confused, not less
<VITAS[m]>
every kind of error in ats has its own error page
<VITAS[m]>
apache webserver on the sd hosts has its own too
<VITAS[m]>
there are instances where ats works and reaches apache webserver but that in turn doesnt reach gunicorn
<DasSkelett[m]>
And one of these error pages is shown on planned maintenances. Just edit this single error page to show the downtime information.
<VITAS[m]>
errors can be different depending on the reason for the downtime
<VITAS[m]>
it can as well be unintended
<VITAS[m]>
or me working on ats
<DasSkelett[m]>
But these outages won't be planned beforehand, so we don't have any information to give.
<VITAS[m]>
no
<DasSkelett[m]>
I think you are making this artificially way more complex that it has to be.
<VITAS[m]>
but even planned ones have different errors
<VITAS[m]>
depending on the work i do
<VITAS[m]>
at the time
<VITAS[m]>
i think this discussion is not needed
<VITAS[m]>
ill make a page myself
<DasSkelett[m]>
You don't think any user cares why the page is down, do you?
<VITAS[m]>
yes they care
<HebaruSan[m]>
Users care about two things, when will it be back up, and (if they're a very conscientous user) should I report this problem to VITAS
<VITAS[m]>
just watch the forums when the site is down
<VITAS[m]>
ill solve it myself np :)
<DasSkelett[m]>
They want to know when the site is back up, not whether it is the ATS, the web server, gunicorn, Flask, SqlAlchemy, the database or whatever.
<VITAS[m]>
yes
<VITAS[m]>
and i want to tell them: this is intentional and you can expect it to be back by x or this isnt intentional you should nudge me
<VITAS[m]>
not all errors are caused because the whole site is down but they acan as well be encountered when it is
<VITAS[m]>
i will make a site put the info on it and link it in the error page
<VITAS[m]>
i want people to be well informed and calm
<HebaruSan[m]>
I'm going to post on the CKAN thread, since those users will just see failed downloads in CKAN during the downtime
<VITAS[m]>
good idea
<HebaruSan[m]>
How long was the downtime going to be, maximally? Was it six hours?
<VITAS[m]>
dc said 800-1600 central europe
<HebaruSan[m]>
OK, so 8 hours, cool
<VITAS[m]>
my guess ist up to 1h in the morning when i update everything and then a hopfully small hickup arround noon
<VITAS[m]>
its a good test if the storage comes back with the new setup
<VITAS[m]>
so site will be there but now dl or ul
<VITAS[m]>
-w
<VITAS[m]>
for like 10mins or if it messes everything up 1h
<VITAS[m]>
i wont leave the site off for 8h
<VITAS[m]>
reminds me: i need a ro manual switch for such instances
<VITAS[m]>
not good to take storage away when someone is uploading
<VITAS[m]>
good
<VITAS[m]>
can you post that as reminder in the sd thread too?
<HebaruSan[m]>
Will do
<VITAS[m]>
(i allready announced it a few days ago)
<VITAS[m]>
so we can test auto gamever addition to sd with 1.10 then?
<HebaruSan[m]>
That's not merged yet
<VITAS[m]>
k
<VITAS[m]>
so ill add it manualy then
<HebaruSan[m]>
Submitted it after we were already in soft feature freeze
<VITAS[m]>
k no problem :)
<HebaruSan[m]>
OK, now whoever freaks out first and rushes to the CKAN or SD forum threads to ask what is going on, will be faced with an announcement already explaining it
<VITAS[m]>
my checklist says ive to do config.inis on the new prod hosts and have done everything i can think off after that
<DasSkelett[m]>
...which they will ignore and ask what's up anyway.
<VITAS[m]>
yes
<VITAS[m]>
i know
<VITAS[m]>
your duty is to explain then :D
<VITAS[m]>
ill be bussy doing the server dance
<VITAS[m]>
will be fine :)
<VITAS[m]>
better planed than most downtimes :P
<HebaruSan[m]>
In reality, only non-redistributable mods should really be affected for CKAN, since we fallback to archive.org for the rest. That should give us some breathing room.
bt_dang[m] has quit [Quit: Idle timeout reached: 10800s]
DasSkelett[m] has quit [Quit: Idle timeout reached: 10800s]
VITAS[m] has quit [Quit: Idle timeout reached: 10800s]
HebaruSan[m] has quit [Quit: Idle timeout reached: 10800s]
Darklight[m] has joined #spacedock
<Darklight[m]>
I'm going to be at work so won't be able to help for the update