Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[wierd]

So, you have a bandwidth intensive application, that you want to live within a specific network.

Do you intend to have that network completely isolated from the network, or would some routing rules be sufficient? (we can configure that subnet to have no route to the internet. Only a route to the internet-enabled subnet.)


also, I need to know the exact models/brands of router you have acquired.



General idea:


OBS network:

192.168.2.X
Has a path in the routing table for that subnet, for how to get to 192.168.1.X, but no "all other networks" route. this means it can see, and communicate with your other network but not the internet.

Main windows network:

192.168.1.X
Has a route to the OBS subnet, and a route to the "all other networks" (aka, the internet).



Nitty gritty:

The OBS network is configured with a static IP address on its WAN port. (the one that normally plugs into an internet modem of some kind.) That number is a 192.168.1.X type number.  (in this case, 192.168.1.254) The WAN port has a patch cable connecting it to an arbitrary port on the main router.

It is configured to dole out DHCP addresses in the 192.168.2.X subnet.  Devices connected to its main ports will get addresses in that range.

It has a routing table that looks like this:

Code: [Select]


Network destination Netmask         Gateway Interface Metric
0.0.0.0         0.0.0.0         127.0.0.1       127.0.0.1 10                <----This is the "All other routes" route.  We point it at the reserved "Myself" address. This blackholes all other networks as 'undeliverable.'
192.168.1.0         255.255.255.0         192.168.2.1 192.168.1.254 10                <----This is the route that tells traffic outbound to the internet enabled network what interface to get to that network over. This is the hard set IP for the WAN port
192.168.2.0         255.255.255.0         192.168.2.1 192.168.2.1 10                <----This is the route that says all traffic destined inside the 192.168.2.X network should use the router's internal network address (hard set) as the routing number



The main router has a routing table that looks like this:

Code: [Select]

Network destination Netmask         Gateway      Interface         Metric
0.0.0.0         0.0.0.0           192.168.1.1        [WAN IP]         10                <----This is the "All other routes" route.  This will be set by the DHCP functionality of the internet enabled router, as it negotiates with the ISP.
192.168.1.0         255.255.255.0           192.168.1.1      192.168.1.1 10                <----This is the default route for within the subnet. Says to use the router's IP for all local traffic routing tasks.
192.168.2.0         255.255.255.0           192.168.1.254      192.168.1.254 10                <----This is the route to get to the 192.168.2.X network. Says to forward all traffic intended for that network to that that local IP address. This is the statically assigned IP of the WAN port of the other router.



The instructions for how to set up these manual routes will be specific for that model of router.

[dragdeler]

Ooh. Now that you say it, the computer connected to both networks does have 192.168.1 IP on both networks, as if they were the same. Both routers are trying to be 192.168.1.1 right? Aaaand yup if I disconnect the wifi on 192.168.1.1 now the password on the label from the router next to me works.


They're both b-box3, is that even safe to disclose? Uhm other than portmapping, only options I find where I can type in numbers is "DHCP" and "DynDNS". The portmapper looks like this.

Spoiler (click to show/hide)

I'm kinda tempted to unplug wifi again, go in there and just set DHCP to 192.168.2.1... Is that a possibility or need I find a way to translate your table into the portmapper?

[wierd]

Its a routing table. Not a portmap.

The routing table is how the router routes packets between networks.

Sadly, it looks like those boxes are locked up tight. No way to do a custom route.

in which case, the best option is to have one box be a .2.x one, the other a .1.x one, and just have a machine with 2 NICs.

[dragdeler]

So the DHCP setting it is then?

I'll keep my eyes open for a better router,  or even just a simpler one, I figured proximus is not supplying people with the good tech. People throw away perfectly usable stuff all the time, had I known I needed one I allready saw a bunch pass me by.



I never tried and I don't have one here, but I suppose those small little "generic" routers like in the picture all host a page with their settings too right?

Spoiler (click to show/hide)

[wierd]

Ideally, i would get one with ddwrt or openwrt support. (Check both projects for supported hardware)

https://wiki.dd-wrt.com/wiki/index.php/Supported_Devices

https://openwrt.org/toh/start

Ddwrt is better for novices.

Openwrt for real admins who like to administer over ssh.


Either will give you access to all the things.

I would look for a gigabit or faster router, with 64mb (or larger) flashrom, and 512m or larger ram, supported br ddwrt. (In your case. It has a better gui management portal)

Adding the custom route would be very straightforward.

[dragdeler]

Damn. I think I have everything I need there, I still haven't totally processed the sheer density of information, but thanks a lot for your help. Might as well do it right if I do it.

[wierd]

No rush.

Do it at your leisure. 


For now, just set one to be active in the 192.168.2.X space, and use your OBS capture stuff on it.  It wont be able to see the other network/router without the custom routes, so dont bother linking them together.  When/If you get suitable network gear that actually gives you real routing capabilities, hit me up, and I will help you set it up.

[chaotic skies]

Maybe not the best place for this, but here goes-

I have a set of data and a value. I believe that value to be the sum of a subset of that data, with the only apparent unique characteristic of that subset being that it is part of my sum. Is there a way to find that subset without brute-forcing it? Is there a way to brute-force it for small quantities?

[LordBaal]

Excell? Access? SQL? Java? C?

[chaotic skies]

Currently have it in Excel, it's not a big data set but I need some way other than doing this by hand.

[LordBaal]

You can use solver.

Fake edit: found a little guide: https://thedailycpa.com/identifying-excel-entries-that-add-up-to-a-specific-value/

[chaotic skies]

Thank you! This is exactly what I was looking for

[Starver]

Though this essentially is solved, programatically I would do something similar but without the limitations and yet possibly more optimal.

Code: (ad hoc pseudocode, obviously untested) [Select]

  set List as values, sorted by value, of structure "value, List of instance contexts"
  // Could be overkill, maybe we can ask the user to resolve the values back to context themselves
  // But this way we also can shortcut things if multiple contexts have identical values
  // ...especially if there a multiple of multiple-contexts
  // PS: assuming all values are positive non-zero! If it's otherwise, still possible but slightly more complicated...
  set Target = whatever // natch.
  set Progress = empty // keeps track of progress, initiated 'clean'
  set Answers = empty // Optional... Could just fall out with the very first identifiable success, but...

  Seek(List,Target,Progress) // as defined below

  define Seek(InList,CurrentTarget,CurrentProgress)
    if InList.count<1 then return // One of several possible checks for having hit an empty loop
    // Although, without it, it'll just not do any the following anyway. Still, does not harm to check!
    while(InList has elements)
      TryElement = pop InList // Shave down (this copy of) the list, each time
      // See how many of this value is possible
      for OptionCount = 1 to TryElement.contextList.count // Start with assumption it can be any/all of this value's contexts
        ThisValMult = TryElement.value*OptionCount
        if ThisValMult > CurrentTarget then next while // Bust! This element can no longer play a part in any answer
        // But if not bust then possible
        OnwardProgress = CurrentProgress + Summarise(OptionCount,TryElement.contextList)
        // However you want to say/store it. If there's just one list item (e.g. "Photocopier Repair Costs"), that's one thing.
        // But perhaps could be "Any 3 of the 4 contexts: 'Van Hire', 'Aircon Repair', 'Bob's Birthday Party', 'Given to hobo'" ;)
        // I leave you to write the Summarise() function yourself, according to your preferences/sado-masochism with data!
        OnwardTarget = CurrentTarget - ThisValMult
        if OnwardTarget == 0 then
           Answers = Answers + OnwardProgress // Or just quit with 'the' answer
           next while // But if we don't why not hunt for the next possible answer(-combo)??
        endif
        Seek(InList,OnwardTarget,OnwardProgress) // Having not failed *or* satisfied the answer, reiterate onward
      end for // end of OptionCount loop all possible context counts
    end while // end of steadily decreasing TryElement candidates
  end define Seek

  Display(Answers) // Depends on how you stored it, and want to present the stored.
  // You could always use this to function to extract the *best* answer(s) from all possibilities.
  // Least/most elements needed, perhaps. Up to you and your needs.
  // If you didn't just display the first achieved target you found!

Anyway, could be a sledgehammer to crack a nut. But vs a potentially 2^N checks of N items (only to eventually fail, or give you the 'best' miss, if there's actually no valid answer) it shortcuts the potentially full-fat bruteforced effort by trimming its own tree and treating essentially option-branches the same (and telling you where you are left with those options) and making sure you know where that leaves you.

There's a breadth-first option, too (build up an OnwardProgress list, each entry tied to its own OnwardTarget and remaining InList options, of course, or reconstructing on each further-inward iteration) but that'd only be best if you want to just seek the minimal 'collection' (or an awkward recursion limit) and have no worries about exponential growth of parallel dendritical search-feelers. Not otherwise recommended, but it's a design option.


That is superfluous to needs, probably.

[bloop_bleep]

Maybe not the best place for this, but here goes-

I have a set of data and a value. I believe that value to be the sum of a subset of that data, with the only apparent unique characteristic of that subset being that it is part of my sum. Is there a way to find that subset without brute-forcing it? Is there a way to brute-force it for small quantities?

This is the subset-sum problem. It is NP-complete, i.e. difficult to do quickly.

[Starver]

Interesting development over on Chrome for Android[1], re: the current incarnation of page-tabs. Wondering if anyone else has seen this.

A couple of years back, the top-of-window 'tabs' (very handily rearrangable) became a drop-downable tab-list (not rearrangable, and initially "new 'tabs' to the bottom", until probably people complained and they made it make it top-of-list addition. (The obvious drag-to-re-order method, often employed, does not work. Closing and recreating a tab shoves it to the 'new'/top-end, but that's a fudge.)

Maybe a year back, they introduced 'tab groups', which essentially filled a single tab-list slot with a series of pages that could be accessed, within thst group, by bottom-of-window (transient, you sometimes had to fiddle a bit) string of tabs-in-this-group. Almost like tabs used to be (still no re-ordering!), although jumping into a group from other top-level tab(groups) in the big vertical tab-list page present a secondary vertical tab-list for which sub-group member.

What with having various long-term tabs (of whatever era) just hanging around - though not as many as I have here in Firefox! - it's messy, but you learn to live with it. The group-tab switch is also useful to bring up the onscreen keyboard when it refuses to[2].

(And, long after that last update that rejigged everything, it started to flash up a highlight of the button used to do top-level tab-switching with a hint-bubble to tell me what it's for ("Open tabs to visit several pages at the same time."), like I'm a first-time user. I think some internal 'intro' flag has been set in a way that nothing I do proves that it can now be unset. Really awful change-control by the developers...)


So now suddenly, and without apparent app-update (not even silently, SFAICT) the tabgroup situation has vanished, all its members folded intothe main tablist. The first in the group is now first in the list (not sure if that's because I created it anew, while trying to work out what was going wrong), the rest every tabgroups' members now sit (ungrouped) between the ungrouped tabs, in the gaps where the groups they cam from were. Long-hold menu still features the "Open in new tab in group" option as well as "Open in new tab", but whether that happens (and the result just appears flattened) or it now has no groups go to, I haven't yet decided.


More testing, later.  For now, I just want to record my situation somewhere, and the official fora/etc are godawful, I've never been able to get on with them, so you lucky people get first refusal on telling me that I'm the one that's doing it wrong. Probably in so many ways! But you know how things are...


Addendum: Adding a new tab (once again) sends it to the bottom of the list again, so the first group-tab tab that made it to the top of the master-tablist was clearly its own idea, not my messing about trying to understand things.


[1] Almost exclusively used for Wiki editing and any incidental jump-off links I follow.

[2] Sometimes, when putting the focus into a textarea edit box (and only in Chrome) the onscreen keyboard does not pop into existence, however long one waits. Cut/Copy/Paste works, if that's all I need, but to unstick it 'properly' I can change tabs (within subgroup or not) and return and hey-presto it materialises for use. That's definitely not an intended operational feature. But you get used to it...