Joined: 10 Nov 2018 Posts: 350 Location: South Carolina
Posted: Fri Apr 15, 2022 7:55 Post subject: Not so random disconnects, Not reboots. wrt3200acm v1
Folks,
I've been beating my head on the table over a problem I have with disconnects on a "Wired" network. Wifi is NOT involved. Firmware versions seems irrelevant since r48646 back to r48261. My gut is pointing a finger at either a setting, or a soon to be router hardware failure, for this has only began happening in the last month or so. I don't believe its a general firmware issue, for no one else seems to report the same situation. Connection is fiber 1000/500 - No ISP devices between the router and the modem. I'm in full control of my network including the router firewall which is obviously on. No DMZ or port routing going on.
The basic problem is a LAN associated disconnect to every device on the network. Happens randomly (I think), and not every device disconnects simultaneously. One device will drop, followed by a couple others, eventually all devices, then they all re-connect. All happens within about 30 seconds. The syslog seems meaningless, else I would post it here. The router is NOT rebooting. Happens a few mins between failures, to more than several hours between failures. Doesn't seem to have a time period between failures. Connections are both internet and intranet only connections. For example video streaming over the wan, as well as p2p connections inside the intranet.
Since the wrt3200acm only has 4 ports on it, I am splitting one port via a bridge (switch) into 7 additional hard wired devices. So the router is supporting a total of 10 wired devices, and maybe a few wifi connections which are NOT always up. Accordion to memory displays and processor usage, I never see more than about 5% to 10% activity, so it doesn't seem to make sense that I would be stressing the router, for not all devices are traffic heavy, and displays imply it can handle much more.
Questions I have to perhaps get the discussion going. Would/could a packet buffer overflow of some kind "crash" the system, causing a disconnect and reconnect? Can a failing application which dumps a lan connection, cause the router to vomit on itself and disconnect everything only to almost immediately recover connections without rebooting?
I'm going to throw this in the air, and see what you folks think. Are there settings I can modify to help out? If so, what? Or, what else do you think could be going on?
Last edited by scar1943 on Fri Apr 15, 2022 13:52; edited 1 time in total
Joined: 04 Aug 2018 Posts: 1402 Location: Appalachian mountains, USA
Posted: Fri Apr 15, 2022 17:01 Post subject:
My EE instincts are suspicious that a hardware failure is currently intermittent, possibly heading toward complete failure, in the router's internal ethernet switch. Might be interesting to add some extra fan action to the router to keep the electronics as cool as reasonably feasible, to see if it changes anything.
Do you ever lose internet? Switch failure that affects the LAN ports would likely affect the Internet/WAN port as well. Of course a failure could also be upstream of the switch. _________________ Netgear XR500 on 52369, 4x Linksys WRT1900ACSv2 on 51530: VLANs, VAPs, NAS, station mode, OpenVPN client (AirVPN), wireguard server (AirVPN port forward) and clients (AzireVPN, AirVPN, private), 3 DNSCrypt providers via VPN.
Joined: 10 Nov 2018 Posts: 350 Location: South Carolina
Posted: Fri Apr 15, 2022 17:56 Post subject:
kernel-panic69 wrote:
Have you checked all of the ethernet cables? What OS are these wired clients running? How often are your client releases set to be renewed? Do you have IPv6 enabled?
Hi Panic,
Checked all cables? No, but have swapped one of them out to test for changes, none observed and didn't make sense that they would all be bad, besides, they are permanently installed within my homes walls and runners. It would be a major installation undergoing to replace all the cabling. The OS's are windows 10, 11, linux and IOS devices. The DHCP expire is set for 1080 mins, IPv6 is disabled.
SurprisedItWorks wrote:
My EE instincts are suspicious that a hardware failure is currently intermittent, possibly heading toward complete failure, in the router's internal ethernet switch. Might be interesting to add some extra fan action to the router to keep the electronics as cool as reasonably feasible, to see if it changes anything.
Do you ever lose internet? Switch failure that affects the LAN ports would likely affect the Internet/WAN port as well. Of course a failure could also be upstream of the switch.
Hi Surprised,
My router has an added fan to the top, installed it long ago when I noticed, or felt like, the temp display in ddwrt showed it to be what I considered above normal temp. We do lose internet connections occasionally, especially when there are storms local, but not in all these cases. I agree with your statement, except that even the local intranet P2P connections that don't touch the internet also drop if they go into the wrt3200. As far as a hardware failure, I'm hoping not, but have considered it. I was hoping it was more about something I set up wrong or could fix with a setting vs a hardware failure.
Per Yngve Berg wrote:
Do devices that are connected to the secondary switch also lose communication with each other?
No, if I understand your question correctly, any P2P device that talks local through the added 7 port switch, stays up. In other words, if the connection does not need to touch the wrt3200, those connections seem to stay up. In my case, its security camera traffic from a dvr, to a client inside the 7 ports on the switch.
Joined: 08 May 2018 Posts: 13445 Location: Texas, USA
Posted: Fri Apr 15, 2022 20:29 Post subject:
NTP sync cycles certain other network services and related processes (like SFE). You can adjust the NTP interval via nvram variable manipulation:
Code:
nvram set ntp_timer=86400
nvram commit
reboot
The above code will only sync NTP once per day, which should alleviate this issue as it is probably restarting SFE whether it passes or fails to sync. That is IF this is why you have connection drops. I haven't seen this particular issue very often, but it's possible that the dhcp requests, acks, etc. all tend to coincide on this as well. I think I have my dhcp set to once per day here since there's only so many wifi clients whitelisted to be able to connect to my AP. Sometimes, paranoia settings have their caveats (?)...
Joined: 10 Nov 2018 Posts: 350 Location: South Carolina
Posted: Fri Apr 15, 2022 22:19 Post subject:
kernel-panic69 wrote:
NTP sync cycles certain other network services and related processes (like SFE). You can adjust the NTP interval via nvram variable manipulation:
Code:
nvram set ntp_timer=86400
nvram commit
reboot
The above code will only sync NTP once per day, which should alleviate this issue as it is probably restarting SFE whether it passes or fails to sync. That is IF this is why you have connection drops. I haven't seen this particular issue very often, but it's possible that the dhcp requests, acks, etc. all tend to coincide on this as well. I think I have my dhcp set to once per day here since there's only so many wifi clients whitelisted to be able to connect to my AP. Sometimes, paranoia settings have their caveats (?)...
Sorta' seems strange that would cause that. I'll give it a try.
Something I'm pondering, can an unrelated connection severing cause the router to dump all the other connections whether connected online or not? For example, there is a p2p connection going from a device to another which is also within the intranet, be dumped if an unrelated connection from the internet is severed for an unknown reason?
