This message was deleted.

Awaiting data entry for an indeterminate amount of time in a web app is a bit of a can of worms (entangled with session timeouts, etc.)

yeah it's tricky 😞

can you describe the workflow case, I’m a bit fuzzy on that

we have a few patch plans that do a lot and sometimes we want to halt a plan, do manual verification and continue, or enter a specific version number that should be deployed somewhere

ahh, so a manual review step

so at the moment we have out::message like 'review monitoring, if stuff is working, write YES into /tmp/something on $primary'

so there’s some thing you just won’t know until part of the plan has run

yes

gotcha

we considered splitting it up. but the first part generates some state that needs to be used in the second part, like "I patched packages Z,X,Y on nodes A,B,C. D,F,G rebooted"

yeah, tricky

if this would be reengineerend, it would probably be split up into multiple plans somehow. but this was initially ported from bolt to PE and it's always tricky to explain to customers why an open source feature is missing in the enterprise version. usually it's the other way around

understood

of course on a CLI this is way easier to implement compared to the Web UI magic

yep, but also riskier 😉

yeah 😄

I’ll roll that around a bit, kinda unfortunate that relay is EOL because that’s excactly the sort of thing it was good at

mhm right