| |
to alpovs
Re: GigaHub chokes with many concurrent network connections.said by alpovs:The PON Serial Number must correspond to what an installer entered into the system and this number can't be spoofed on another modem. Oh yeah it can. There's a whole community around XGS-PON spoofing. |
|
| Nabeel_co |
to alpovs
said by alpovs:It has always been like this with fiber. Back in the HH3K days, you could simply take the SFP+ module out of your HH3K and install it into your own equipment with the slight catch that the SFP+ modules that Bell used were designed to not pull the sense pin down to ground as you're supposed to with SFP, so some devices would never detect anything got plugged in... a simple solder blob over a resistor though... and it would behave like any SFP module and work fine. This is what I did before the HH4K and the GigaHubs. |
|
|
| Nabeel_co |
to Cloneman
said by Cloneman:Should basically be a consumer right to request specific hardware, even if it's at an additional cost. Yeah, I'm basically thinking about wasting their time over and over until either my bypass XGS-PON SFP+ module arrives, or they fix it. I don't care which one happens first... just so long as it happens. This is bullshit. I've notice, also, that this happens anytime I restore my previous session in Firefox as well, because I have a few hundred web pages open, and them all loading at once exceeds that 1000 connection "limit". |
|
| |
to Nabeel_co
I don't think I currently have this problem on my HH4000. I Could benchmark it, and I have access to someone with a gigahub as well (albeit with 500mbit service only in his case)
what's the best way to reproduce the issue? torrent with multiple files? some clients have a connection limit |
|
|
| |
to Nabeel_co
said by Nabeel_co:I've notice, also, that this happens anytime I restore my previous session in Firefox as well, because I have a few hundred web pages open, and them all loading at once exceeds that 1000 connection "limit". That must be a you using PPPoE on your own hardware issue? My chrome or Firefox session has 205 tabs that open and I've never once had an issue with it killing my connection, between when I had the HH4000 and gigahub for the last 9 ish months. |
|
paolo
join:2018-11-06
canada |
paolo
Member
2024-May-28 10:05 am
why the heck does someone need "1000 concurrent connections"? what happened to turning on the computer, and opening the app you need like the email app or browser app and doing a simple task like that? why u gotta complicate things? |
|
| |
said by paolo:why the heck does someone need "1000 concurrent connections"? what happened to turning on the computer, and opening the app you need like the email app or browser app and doing a simple task like that? why u gotta complicate things? Why are you responding to me? I never said anything about 1000 concurrent connections. That being said my computer is never shut off. Except for 10-14 days every 2 or 3 years when I go to Disney. |
|
| |
to Nabeel_co
said by Nabeel_co:said by Jelllo:I doubt it is the GH. I have the 8 gig profile and with usenet have over 10 million connections in a minute without a problem. I highly doubt this is actually the case. You're likely misinterpreting the data. Also to reproduce what I've described you need to be using PPPoE passthrough. Also, 10 million active connections would bring most enterprise grade hardware to it's knees, let alone low rent crappy hardware from an ISP. No, you don't need PPPoE passthrough to see this effect. A lot of connections through the GigaHub in router mode without another router will have the same effect. |
|
| alpovs |
to Nabeel_co
said by Nabeel_co:said by alpovs:The PON Serial Number must correspond to what an installer entered into the system and this number can't be spoofed on another modem. Oh yeah it can. There's a whole community around XGS-PON spoofing. Reading comprehension? The discussion was about using modems off Kjiji and the quote: "this number can't be spoofed on another modem". No, you can't spoof PON SN on another modem. And yes, I know about bypassing with an ONT stick where this number can entered and it will work. |
|
| alpovs |
to Nabeel_co
said by Nabeel_co:said by alpovs:It has always been like this with fiber. Back in the HH3K days, you could simply take the SFP+ module out of your HH3K and install it into your own equipment with the slight catch that the SFP+ modules that Bell used were designed to not pull the sense pin down to ground as you're supposed to with SFP, so some devices would never detect anything got plugged in... a simple solder blob over a resistor though... and it would behave like any SFP module and work fine. This is what I did before the HH4K and the GigaHubs. Again, the discussion was about using non-Bell or aftermarket modems. What you described is well known and very different from using aftermarket modems. |
|
| alpovs |
to Nabeel_co
said by Nabeel_co:Yeah, I'm basically thinking about wasting their time over and over until either my bypass XGS-PON SFP+ module arrives, or they fix it. They will never fix unless they move to another modem/ONT. This issue is a property of the GigaHub. |
|
| alpovs |
to Cloneman
said by Cloneman:1. - Cellpipe Modem (first VDSL Modem)
"Sync No Surf" Issues
2. - Sagemcom Modem 2864
Modem Crash when using too many connections
In both cases a modem replacement from the grey market solved the problem, whereas the official policy from bell (for Quebec) was you would not get a HH2000 modem until you had entertained 3 technicians in your home, who would of course fail to find a solution because they were replacing the same modem with the same broken firmware.
Most recent Bell problems that I've seen over can be traced back to them refusing to put you on a working DSLAM or put you on a working modem.
Without the ability to swap in a modem, you have limited troubleshooting ability.
Should basically be a consumer right to request specific hardware, even if it's at an additional cost.
They spend billions of dollars on a robust network only to have their customers & staff handcuffed by hardware policies. I am not sure how this is relevant to what you said when this discussion started: said by Cloneman:they have done a very poor job of recalling rental modems for return. Sending rental modems for return has always been easy. Supplied equipment has always been bad, yes. I am surprised you didn't buy a SmartRG or Huawei modem those times. I bought a SmartRG modem from an independent ISP when VDSL became available. I used it with other VDSL providers until I subscribed to 100/10 VDSL from Bell when I had to use their modem. |
|
alpovs
1 recommendation |
to paolo
said by paolo:why the heck does someone need "1000 concurrent connections"? what happened to turning on the computer, and opening the app you need like the email app or browser app and doing a simple task like that? why u gotta complicate things? Because families have more than one person and a few young people who do more than opening email or browser. |
|
paolo
join:2018-11-06
canada |
paolo
Member
2024-May-28 12:57 pm
said by alpovs:Because families have more than one person and a few young people who do more than opening email or browser. I get that, but 1000+ connections? that's probably more of a WANT than a NEED? ya think? |
|
2 recommendations |
said by paolo:I get that, but 1000+ connections? that's probably more of a WANT than a NEED? ya think? So are we safe to assume you didn't know that even going to a single website isn't 1 connection. |
|
| |
to Nabeel_co
said by Nabeel_co:said by Cloneman:but maybe the slave mindset on forums can. A bypass device isn't practical for everyone, and in my case, I will be bypassing my GigaHub, but even with an order in for my own XGS-PON, it may take months to actually arrive. I bypassed with the WAG-D20 and currently with the WAS-110, the issue still occurs on my end. |
|
| |
to paolo
said by paolo:I get that, but 1000+ connections? that's probably more of a WANT than a NEED? ya think? Can you check in your router how many connections you have now? I have 745 at the moment and not everybody in the family fired up their internet devices yet. I think you are misinterpreting the term connection or state. Run 'netstat -a' on you PC and surprise yourself. |
|
alpovs
1 edit |
to djbuddha
said by djbuddha:I bypassed with the WAG-D20 and currently with the WAS-110, the issue still occurs on my end. It must be a different issue then. You can join the Discord server and seek help in finding out what the issue is. |
|
1 recommendation |
to Cloneman
said by Cloneman:what's the best way to reproduce the issue? It happens for me, simply by connecting to a dedicated server in the Steam version of Valheim, because it creates a few thousand connections for some reason in rapid succession. BUT, it seems like you have to be using PPPoE Passthrough. Of course, I'd say just throw it all at it, Valheim, Torrents, etc, just to make it happen. If you have a way of monitoring your NAT table on your router, you should be able to see how many active connections are going at once, and that should let you see if you're getting close to C1K. Also run a ping test while this is happening, and you'll see the pings skyrocket. |
|
| Nabeel_co |
to Artwebb1986
said by Artwebb1986:That must be a you using PPPoE on your own hardware issue? Nope, my hardware can handle over 1 million concurrent connections, and CPU usage is less than 1% when all this is going on. The GigaHub however will max out it's CPU and start to lag if you have other devices on it directly. It's only the GH that starts to lag because the PPPoE passthrough is actually handled by the GH. It emulates a PPPoE server, and de-encapsulates your PPPoE traffic and re-encapsulates it through it's own second PPPoE link to bells servers, so at any one time it has to maintain 2x a connection list of active connections plus whatever connections it's running itself on it's own PPPoE link. |
|
| Nabeel_co |
to paolo
said by paolo:why the heck does someone need "1000 concurrent connections"? That's not hard to do... Like I've said in the past, some games will establish up to 5000 concurrent connections just to check for other players on the internet. 1000 active connections is nothing. Most low end hardware can do 10,000 easily. My router can handle over 1 million concurrent connections. Just for frame of reference, in the late 90s, the C10K problem was a thing, which was a single device handling 10,000 connections, and it was solved by the early 2000s. It isn't normal for modern networking hardware to have issues with anything less than, I'd say, 100,000 concurrent connections. |
|
Nabeel_co
1 recommendation |
to paolo
said by paolo:I get that, but 1000+ connections? that's probably more of a WANT than a NEED? ya think? No, that's literally nothing. For some context, Facebook opens about 280 connections when load the page, Twitter about 208, NYTimes about 170, etc... Have 5 people browsing Facebook, and your network could screech to a halt. And again, as I've mentioned before, many video games on Steam open up to 5000 concurrent connections to do things like load a list of servers or players. 1000 connections only sounds like a lot until you realize most apps, web pages, etc, open about 200 on their own. |
|
| Nabeel_co |
to djbuddha
said by djbuddha:I bypassed with the WAG-D20 and currently with the WAS-110, the issue still occurs on my end. Oh that's interesting... Are you sure your MTU values are set correctly? From bypassing my HH3k, I remember that the HH with PPPoE passthrough was far more forgiving of bad MTU settings than with the bypass, because you'll need to do things like setting up MSS clamping properly and whatnot as well. I can't see what else it would be, because the GigaHub is DEFINITELY choking and dying. so if it's still happening, then that means there's a new and different problem down stream. |
|
| |
to Nabeel_co
said by Nabeel_co:Nope, my hardware can handle over 1 million concurrent connections, and CPU usage is less than 1% when all this is going on. The GigaHub however will max out it's CPU and start to lag if you have other devices on it directly.
It's only the GH that starts to lag because the PPPoE passthrough is actually handled by the GH. It emulates a PPPoE server, and de-encapsulates your PPPoE traffic and re-encapsulates it through it's own second PPPoE link to bells servers, so at any one time it has to maintain 2x a connection list of active connections plus whatever connections it's running itself on it's own PPPoE link. Yet I've used only the gigahub and never had a slow down since day 1. |
|
| |
said by Artwebb1986:Yet I've used only the gigahub and never had a slow down since day 1. Are you using PPPoE passthrough? How are you monitoring the number of active connections? What's your network setup? |
|
| |
said by Nabeel_co:Are you using PPPoE passthrough? How are you monitoring the number of active connections?
What's your network setup? As I've said I just use the gigahub. Not monitoring connections at all because it doesn't matter haven't had a single issue since day 1. 10gig to my PC, gigabit to my Nas, to an old 8 port switch behind my tv to wire up the pvr, nintendo switch and odroid n2+. |
|
| |
Yeah, so again, as I've mentioned before, this happens, likely because of PPPoE passthrough, because each connection is at minimum doubled, and it results in 3 PPPoE connections being managed by the GigaHub.
For those of us who don't want to rely on the GigaHub managing our network because if its serious limitations, we do PPPoE passthrough. |
|
| |
to Nabeel_co
said by Nabeel_co:It's only the GH that starts to lag because the PPPoE passthrough is actually handled by the GH. It emulates a PPPoE server, and de-encapsulates your PPPoE traffic and re-encapsulates it through it's own second PPPoE link to bells servers, so at any one time it has to maintain 2x a connection list of active connections plus whatever connections it's running itself on it's own PPPoE link. Really? Emulates a PPPoE server? Where did you get this information? Any links for reference? |
|
| alpovs |
to Nabeel_co
said by Nabeel_co:Yeah, so again, as I've mentioned before, this happens, likely because of PPPoE passthrough, because each connection is at minimum doubled, and it results in 3 PPPoE connections being managed by the GigaHub.
For those of us who don't want to rely on the GigaHub managing our network because if its serious limitations, we do PPPoE passthrough. It may sound discomforting to you but I used a GigaHub in PPPoE passthrough mode and never had a slow down or noticeably high latency. said by Nabeel_co:each connection is at minimum doubled, and it results in 3 PPPoE connections Does it make sense? |
|
1 edit |
to Nabeel_co
I tried to reproduce issue today on my HH4000 and was not (yet) successful. using qbittorrent modified to allow 1900 connections, I could not get any packet loss on my various pings being sent to the internet from another computer. packet loss does start to happen when I uncap qBittorrent to use more than 300 - 400mbit, but I think I've discovered that that's a limitation of my router, infamously pinned to 50% sirq on a dual-core architecture above a certain speed. opening 40 tabs in firefox of lage websites at the same time does not produce any ill effects either. Not sure if TCPview is providing an accurate representation of the number of connections I have open, I've seen it go over 1000 "Endpoints", whatever that means. Need a better benchmark test that uses a lot of connections, perhaps 4,000 or more, to get a meaningful test result qbitorrent appears to be capped at 2,000 by the software developer, I woudn't be suprised if firefox has a low limit as well. Perhaps Valheim is sending a large number of packets on small connections, rather than opening many connections. relevant reddit thread: » www.reddit.com/r/valheim ··· s_on_my/Remains to be seen weather this behavior is overwhelming the ISP modem or the personally owned router. I would recommend double checking to make sure the load average is not high on the router during this time to rule this out, because mine was at 1.00. |
|
