Subnet range selector, interface fixes #481

0xCA · 2023-11-06T00:36:41Z

Subnet range selector

10.0.0.1/16 server with a very few clients, correctly suggesting an IP from 10.0.4.0/22 for a new client

Reason

This feature is intended for subdivision of large wireguard servers, such as /16 subnets.
I wanted to allocate smaller blocks of the server address space for specific purposes and easier firewalling.
Current IP suggestion is unaware of such blocks, which requires entering each IP manually.
So I made it aware.

Details

Subnet ranges are set in an optional variable SUBNET_RANGES
Format: SR Name| 10.0.1.0/24; SR2| 10.0.2.0/24, 10.0.3.0/24, 2001:0db8::/32
Each CIDR must be inside one of the server interfaces
Other than that, you can go for a really small CIDRs, no problem
Any invalid CIDRs or entire SRs are discarded during the server load
The list is only stored in memory, so you can change it by changing the env variable and restarting wireguard-ui
Multiple server interfaces or multiple client CIDRs are supported, but not auto-suggested
No part of this feature is saved to the database

IP suggestion

IP is suggested from a chosen SR
If no free IPs are left, an error is shown
Default SR is Any, which is "any server interface", same as no SR at all
For existing clients, you can select a different SR, a new IP is automatically suggested
If you select the same SR, a new IP is suggested
Fix for multiple interfaces: it only fails if no interface has free IPs

Subnet range filter

Allows to show clients from a selected subnet range for easier managing
Pure client side, no requests
A SR is detected server-side when wg client info is requested
Only the first match is used
There is an in-memory lookup cache on the server, that reduces subsequent cidr-contains-ip evaluations

Extra

I also Implemented some interface/UX fixes

Apply config button is properly shown or hidden when needed
Toast notifications are not obstructing the buttons anymore

nebulosa2007 · 2023-11-06T08:36:05Z

You done great work, but you want implement several options at one time and it's not convinient for development..
Could you split you changes on three branches:

ip-range-selector
interface/UX fixes
adding a new makefile (if it new file really needed for options above)

It will increase the chances of appearing in the main branch

0xCA · 2023-11-06T16:26:57Z

@nebulosa2007
It's actually going to be less convenient for development if I split it. One part will conflict with the other, because the changes are in the same places.
The only thing I can split without affecting anything, is input tag fix.
As for Makefile, it's there purely for convenience during the development. I can remove it from the branch, if asked by the maintainer.

systemcrash · 2023-12-03T00:06:30Z

@nebulosa2007 is right here.
-Remove the makefile additions. They're not useful in this repo. (maybe for you to compile locally)
-Separate the UX fixes from the go code fixes for the SR stuff.
-Drop most of the formatting fixes here and there

Makefile

README.md

handler/routes.go

templates/base.html

util/config.go

templates/base.html

util/util.go

0xCA · 2023-12-03T17:53:01Z

-Separate the UX fixes from the go code fixes for the SR stuff.

The SR is not a purely server-side update. I have to change the UI. As for UX fixes, it's not worth the effort separating it, only for the merger to deal with additional conflicts.

-Drop most of the formatting fixes here and there

Is it really worth it? I do understand that it might prevent a minor conflict here and there. But it significantly degrades the DX. OR requires a manual "second pass" with all the formatters turned off (which is not particularly convenient to achieve, let alone the process of manually reverting every fmt change).

0xCA · 2023-12-03T17:54:46Z

#481 (comment)
#481 (comment)
#481 (comment)
@systemcrash
Could you please keep your answers in their threads?

systemcrash · 2023-12-03T17:57:20Z

We would not ask for it if it wasn't worth it. It makes reviewing this and any PR painful when we need to separate formatting from the noise of what's actually important. Rule: Put them in separate commits, in a separate PR. Then it's easy for anyone to understand what the PR is supposed to do. I like the core of this PR, otherwise.

…

Is it really worth it? I do understand that it might prevent a minor conflict here and there. But it significantly degrades the DX. OR requires a manual "second pass" with all the formatters turned off (which is not particularly convenient to achieve, let alone the process of manually reverting every fmt change). — Reply to this email directly, view it on GitHub <#481 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAE6DULTD6D2VORMI4VMAWLYHS4APAVCNFSM6AAAAAA66W2LKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZXGU2TCNZZGI> . You are receiving this because you commented.Message ID: ***@***.***>

systemcrash · 2023-12-03T18:00:00Z

Also, if for some reason anyone needs to revert this PR, they don't need to deal with the fallout of formatting unexpectedly removed which affects commits on top of it also. So dedicate a PR to running go fmt on the code base. 👌

…

Is it really worth it? I do understand that it might prevent a minor conflict here and there. But it significantly degrades the DX. OR requires a manual "second pass" with all the formatters turned off (which is not particularly convenient to achieve, let alone the process of manually reverting every fmt change). — Reply to this email directly, view it on GitHub <#481 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAE6DULTD6D2VORMI4VMAWLYHS4APAVCNFSM6AAAAAA66W2LKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZXGU2TCNZZGI> . You are receiving this because you commented.Message ID: ***@***.***>

systemcrash · 2023-12-03T18:00:45Z

Sorry, on email. Could you please keep your answers in their threads?

…

0xCA · 2023-12-03T18:15:02Z

Sorry, on email.

Are you going to move it later? I can wait and continue later. Or we could continue in the main thread, but I don't think it's a really good idea.

Edit: I just noticed you already moved your answers

0xCA · 2023-12-03T18:36:38Z

So dedicate a PR to running go fmt on the code base. 👌

Sure thing, for an active repo. Not sure what to do here though. Turn off all the formatters for this repo only?..
The more time passes, the worse it gets for anyone trying to contribute.

systemcrash · 2023-12-03T18:40:19Z

So dedicate a PR to running go fmt on the code base. 👌

Sure thing, for an active repo. Not sure what to do here though. Turn off all the formatters for this repo only?.. The more time passes, the worse it gets for anyone trying to contribute.

Agreed on the active repo thing. But prepare the PR and branch such that it could go into our repo 😉

Look at the quality of some of the PRs. Some are messy and a rats nest of dependencies where all commits slipped in and the original idea is lost in noise.

0xCA · 2023-12-03T21:48:15Z

I resolved the issues.
As for undoing gofmt, I probably won't do it again. Vscode absolutely refused to disable it, even though I disabled it in 5 places. So I had to use a dumb text editor.
Next time (Telegram PR?) I will either ask someone to fix it and make a PR/diff for my PR branch, or will never publish my changes in the first place. It's just too much effort to combat gofmt, that is actually trying to make things better.

ngoduykhanh

Thank you @0xCA for the PR and everybody here for your contribution to making the PR better. Overall, the changes look good to me. I've just left one comment. I'm happy to merge it if @0xCA can follow up here and resolve the conflicts.

ngoduykhanh · 2023-12-25T19:57:02Z

templates/base.html

+                            <div class="form-group">
+                                <label for="subnet_ranges" class="control-label">Subnet range</label>
+                                <select id="subnet_ranges" class="select2"
+                                    data-placeholder="Select a subnet range" style="width: 100%;">
+                                </select>
+                            </div>


I think we should hide the subnet range input by default if the SUBNET_RANGES environment variable is not set. Not everyone needs this feature. It will probably confuse new users.
And, the same comment for the dropdown list near the search box.

I actually intend to place this to "Additional configuration" <summary>, along with tg userid setting. To hide it, templates or js must be used, which isn't beautiful. It probably isn't necessary if it's already mostly out of the way?
As for dropdown, it's gonna be just default options if no subnet ranges are defined. No change required.

IDK, it was my personal preference. No worries, I can keep it like this and refactor it later.
Leaving it in the "Additional configuration" doesn't seem nice to me as I think it should be part of the IP allocation section.

Then maybe Endpoint should be? It's an uncommon thing.

Endpoint is a standard configuration supported by WireGuard, isn't it?

It is, I mean uncommon by usage. Very useful sometimes, but most of the time it's just server-clients, with only some clients having endpoint config.

fix: free ip search was stopping after no free ip on the first interface fix: toast obstructing the buttons fix: stuck apply config button

If the same subnet range is selected, a different IP is suggested

- Reverted formatting fixes where possible - Clarified the note about updateSearchList() - Fixed a typo - Made SUBNET_RANGES example yaml-friendly - Removed Makefile

0xCA · 2023-12-27T01:39:55Z

@ngoduykhanh I rebased on the latest master. Let's decide on subnet ranges input and it's ready for merge.

ngoduykhanh · 2023-12-27T08:08:52Z

@0xCA I replied to your comment and merged this PR. Thanks again!

0xCA mentioned this pull request Nov 9, 2023

Fixed tag input being too small and unable to fit a CIDR #483

Merged

0xCA mentioned this pull request Nov 18, 2023

Telegram support #488

Merged

systemcrash suggested changes Dec 3, 2023

View reviewed changes

systemcrash reviewed Dec 3, 2023

View reviewed changes

util/util.go Outdated Show resolved Hide resolved

ngoduykhanh reviewed Dec 25, 2023

View reviewed changes

0xCA added 9 commits December 27, 2023 06:07

Parse and validate subnet ranges

92333a0

Full implementation of subnet ranges for a new client, fixes

53eaab0

fix: free ip search was stopping after no free ip on the first interface fix: toast obstructing the buttons fix: stuck apply config button

Added clients page filter by subnet range

2027b3f

Further Apply config visibility fixes. Fixed tag input being too small.

3a789bf

Added semiauto subnet range changer for existing clients

2000b7f

If the same subnet range is selected, a different IP is suggested

Added SUBNET_RANGES to readme

33db92b

A bit more usable makefile

eaa3799

Revert tag input being too small fix

8678a74

Fixed issues mentioned in the @systemcrash review

9e63301

- Reverted formatting fixes where possible - Clarified the note about updateSearchList() - Fixed a typo - Made SUBNET_RANGES example yaml-friendly - Removed Makefile

0xCA force-pushed the ip-range-selector branch from 01cf84c to 9e63301 Compare December 27, 2023 01:21

Fix duplicated IP suggestions

d12ac44

0xCA mentioned this pull request Dec 27, 2023

Fix duplicated ip suggestions #505

Closed

ngoduykhanh merged commit a9be538 into ngoduykhanh:master Dec 27, 2023

0xCA mentioned this pull request Dec 27, 2023

Fixed tag input being too small and unable to fit a CIDR #509

Merged

nebulosa2007 mentioned this pull request Jan 25, 2024

wireguard more wg servers #540

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Subnet range selector, interface fixes #481

Subnet range selector, interface fixes #481

0xCA commented Nov 6, 2023 •

edited

Loading

nebulosa2007 commented Nov 6, 2023

0xCA commented Nov 6, 2023

systemcrash commented Dec 3, 2023 •

edited

Loading

0xCA commented Dec 3, 2023

0xCA commented Dec 3, 2023

systemcrash commented Dec 3, 2023 via email

systemcrash commented Dec 3, 2023 via email

systemcrash commented Dec 3, 2023 via email

0xCA commented Dec 3, 2023 •

edited

Loading

0xCA commented Dec 3, 2023

systemcrash commented Dec 3, 2023

0xCA commented Dec 3, 2023

ngoduykhanh left a comment

ngoduykhanh Dec 25, 2023

0xCA Dec 27, 2023 •

edited

Loading

ngoduykhanh Dec 27, 2023

0xCA Dec 27, 2023

ngoduykhanh Dec 27, 2023

0xCA Dec 27, 2023

0xCA commented Dec 27, 2023 •

edited

Loading

ngoduykhanh commented Dec 27, 2023

Subnet range selector, interface fixes #481

Subnet range selector, interface fixes #481

Conversation

0xCA commented Nov 6, 2023 • edited Loading

Subnet range selector

Reason

Details

IP suggestion

Subnet range filter

Extra

nebulosa2007 commented Nov 6, 2023

0xCA commented Nov 6, 2023

systemcrash commented Dec 3, 2023 • edited Loading

0xCA commented Dec 3, 2023

0xCA commented Dec 3, 2023

systemcrash commented Dec 3, 2023 via email

systemcrash commented Dec 3, 2023 via email

systemcrash commented Dec 3, 2023 via email

0xCA commented Dec 3, 2023 • edited Loading

0xCA commented Dec 3, 2023

systemcrash commented Dec 3, 2023

0xCA commented Dec 3, 2023

ngoduykhanh left a comment

Choose a reason for hiding this comment

ngoduykhanh Dec 25, 2023

Choose a reason for hiding this comment

0xCA Dec 27, 2023 • edited Loading

Choose a reason for hiding this comment

ngoduykhanh Dec 27, 2023

Choose a reason for hiding this comment

0xCA Dec 27, 2023

Choose a reason for hiding this comment

ngoduykhanh Dec 27, 2023

Choose a reason for hiding this comment

0xCA Dec 27, 2023

Choose a reason for hiding this comment

0xCA commented Dec 27, 2023 • edited Loading

ngoduykhanh commented Dec 27, 2023

0xCA commented Nov 6, 2023 •

edited

Loading

systemcrash commented Dec 3, 2023 •

edited

Loading

0xCA commented Dec 3, 2023 •

edited

Loading

0xCA Dec 27, 2023 •

edited

Loading

0xCA commented Dec 27, 2023 •

edited

Loading