I have deleted and re-installed a couple of my module/app(s) (example: module/webtop, module/nethvoice and so on) and the count up on the module/app goes up even though a module/app is deleted and no longer exists.
For instance, I have installed Nethvoice three times, each time I have deleted the original module/app install before reinstalling. I had some challenges, and I wasn’t sure if it had to do with the installation. I understand it, if you have 2 separate installations of the same module/nethvoice. Then it will count up the next install as module/nethvoiceapp2 and so on. I don’t understand why when deleting module/app1, if module/app1 doesn’t exist anymore, why does NS8 Software Center move up the count on module/nethvoice (or any module) to module/app2?
No, this is the expected behavior. Each new module instance increases the counter because the module instance must be uniquely identified throughout the cluster’s lifetime.
Ok. Forgive me since I wasn’t there when this was decided as an expected behavior. This seems confusing a bit.
Maybe it’s because I don’t know your algorithm to get or format for the unique identification, but It seems like that should be a backend change that should be made and not the frontend? Seems like a brand new, different, unique identification can be added without having to up the count when no module/app1 exists anymore, and you’re reinstalling.
The unique identification can be shown in the app details.
The reason for this question is I can see how the current expected behavior could get messy for enduser. Especially for those who want things to be number in an orderly manner.
Many wonder how it works. Can there be a very simple statment please instead of many links to may different views, leading to many different takes, and many wrong conclusions and admin mistakes?
Second link info was from behavioral observation and probably a quick peek at some part of the code, but for sure didn’t verify it against redis database (for a proper analysis or lab test).
The devs are in a better position to offer a brief but more detailed answer on the internals or technical details.
Thank you for all the links, so I could see the ongoing discussion about module numbering. It helps to understand what has happened in discussions and what has been said.
@Amygos shared this on another post that gave me more insight from the manual about modules.
I believe this helps to understand it.
And it says
Every time a module instance is added to the cluster, the new instance is named as the module itself followed by a progressive number starting from 1. Given a module named myapp, instances will be named myapp1, myapp2, etc.
Modules can be managed using these commands:
add-module <module> <node_id>: install a module on the given node with ID node_id; search for module inside enabled repositories and install latest available version. If module is a image registry URL, just install the module straight from it; this method can be used to install customized images.
remove-module [--no-preserve] <module>: remove an installed module; if --no-preserve is given, erase also all module data
So does that mean that on new install it will number app progressively every time through Software Center. But if I do CLI and use `add-module <node_id> I can adjust the numbering when I load a new one with app1 instead of going to app2? (What does managing modules mean?)
module = app
Incremental numbering also from software center as well as from CLI.
You cannot set the app name but choose which app (module) to install and specify on which node you want to install it. From CLI you can also specify which app version to install if there are more than one available or install from a registry (repository).
I understand that seeing a numerically discontinuous set, like module1, module2, module5, module10, may be unpleasant. For this reason, the UI allows you to set an arbitrary name for each module instance, which is then used for display.
Aside from visual concerns, do you see any other issues?