Question about NS8 (count) behavior when creating new containers or migration

Support
,
1

@developers

Hello,

Maybe someone could answer the following questions about NS8, then I don’t have to find out the following via experiments.

Suppose the NS8 is brand new and no container has yet been created in it:

How does NS8 or the (NethServer7>) NS8 migration tool behave when creating new containers if it finds existing (empty) folders in the path “/home/”, e.g. “/home/mail1” or “/home/samba1” ?

Does the “installer” then simply use these folders or then automatically create folders with further numbered bodies, e.g. “/home/mail2” or “/home/samba2”?

If there were earlier in the NS8, he counts up (even if the folder no longer exists), but what is it like when the container is created for the first time?

And how about counting behavior, if already exist the folder “/home/mail1” and “/home/mail3”, the count would then be used with “/home/mail2” (the gap is used) or with "/home /mail4 "continued?

I ask this (once again) in the context that I am still working on a way that is practicable for myself to put the container data on other drives with the NethServer 7> Neth8 migration. If the behavior or the “high count” is predictable, I can prepare matching mount Binds in advance (binds because the Lost+Found folder does not interfere), but of course that does not help if the container install or the migration tool jumps this path …

Regards yummiweb

Addition

Trying yourself is such a thing, since the results are not clear without knowing the process behind it.

For example, I just wanted to try “Odoo”, for example, that has not yet been in this test installation and not as folders in the path.

Nevertheless, the first instance of “Odoo” with Odoo2 under “/Home/Odoo2” was created. Why does that happen? This is not predictable. After deleting the instance, the next was then “/Home/Odoo3” with “Odoo3”. Is there any Odoo1 instance?

2

Existing home folders matching appX name are not used. Then the internal redis counter is incremented by one.

If it a finds a /home/app1 folder, it fails and increments the internal redis counter to +1 (equals 2). The next add module attempt will use /home/app2.

Numbering is always incrementing. It does not reuse gaps.

Do not recall if add-module command allows you to specify a custom app number. EDIT: it doesn’t

Not sure. Either odoo was installed at some point or app installation failed or was canceled, incrementing internal redis app index.

Is it possible to change the redis app index? don’t know but maybe (with redis-cli).

1 Like
3

Many thanks for your response.

Accordingly, it is not possible to carry out a container installation into an existing folder - and of course it is not possible to prepare a mount of the (to migrate) container data in advance?

What do you suggest? What else would there be so that all container data is not on the same base drive? So far, a mount has always been the means of choice in previous service and application concepts under Linux to achieve this.

Sure, you could create your own nodes for each type of service data (database, files, emails, rarely used archive data) and then mount a suitable data carrier in the whole “/Home” directory. Because for the overarching “Home” folder, the existential examination you described does not apply.

However, this would not only significantly increase the complexity, but would need significantly more storage space (Node system) as well as RAM and CPU overall. A mount doesn’t need anything. And by mount, I can even integrate data from a distant network files system (on which I can not einstall a NS8 Node).

Can the existence check be switched off?

Can a “break” be incorporated into the process of testing/creating the duty path that could be used to integrate your own drives?

Greetings Yummiweb

because of these restrictions more and more desperate.