metronet fiber, internet goes down roughly every 24 hours
-
same 3100. same device
i pulled them directly from var\logs i copied them to a word file. ctrl + f what you wanted to see and pasted it here
read top to bottom
i am really hoping a static IP address from the provider will resolve this
called my isp to see if they could enable the static today. not possible i have to wait until tomorrow at 9am EST :(
-
In the dhcp logs view you can filter by the dhclient process and then just copy/paste them here directly without going through Word (or any other editor).
Some of those logs show the gateway not responding to ARP which probably won't be solved by using a static IP. If you can get one though if will obviously solve any dhcp issues.Steve
-
static ip has been set and active for over 24 hours now. NO issues whatsoever. first time ever with this new internet service.
i will look forward to when Internet Systems Consortium DHCP Server 4.3.6-P1 is updated to 4.4.2 within Pfsense per https://www.isc.org/dhcp/
so all of this is fixed for users.
so to continue using Pfsense i will be paying 10 dollars extra a month until the release..
-
It's dhclient not the dhcp server. It will be in 2.5 when that is released, it isn't yet in 2.5 snaps as they are currently built on 12.0.
Steve
-
As I said (in the wrong thread)....
stephenw10 Netgate Administrator about 16 hours ago
Ok, good news. The binary part of the fix for this is now in 2.4.5 snapshots:
https://github.com/pfsense/FreeBSD-src/commits/RELENG_2_4_5/sbin/dhclient/dhclient.cThe full fix also requires changes to the dhclient script which can be applied via the system patches package. I have briefly tested that and it didn't seem to break anything.
That patch is here: https://redmine.pfsense.org/attachments/download/2682/pfsense-dhclient-script-patch.txt
If you're able to test it we may be able to include it in 2.4.5.
Steve
-
no problem Sir.
at this time i have no way to test as i am locked in a one year agreement with a static WAN ip address. my issue resolved.
not sure i provided you any good information. feel free to lock this thread and work with the other gentlemen if that seems best
-
I'm going to test locally but I can only try to simulate a failed dhcp server. It is definitely a bug that would be very good to squash. I'd love to hear from anyone who is hitting it 'in the field'.
Steve
-
@stephenw10 Hello, I feel like I'm running into this.
Quick overview on my situation:
- Currently running ver 22.05-RELEASE on a 2100. (recently updated)
- Ran over 2 years w/o ever having to reboot appliance on Xfinity internet.
- Recent switch to MetroNet Fiber - Internet goes down every 24-36 hours.
- Has happened 3 times so far, feels like it is reproducing reliably within these time frames.
- Manually releasing WAN / renew WAN action on WAN interface - Restores Internet
Trying to figure out a good path forward:
- Pay $10/month for a static IP (I'd rather not)
- Can we script something to automate the release/renew action during the night on a daily basis? (Feels like a hack)
- Is this a bug that can be fixed? (best solution IMO)
-
What do you see logged when it happens?
The actual bug referenced here was fixed in 2.4.5:
https://redmine.pfsense.org/issues/9267I suggest you may well be hitting this:
https://forum.netgate.com/post/1063443So specifically MetroNets broken DHCP relay behaviour. The workaround shown there should prevent it if so.
Steve
-
I haven't caught logs when the issue happens yet. I'm hoping I can catch on the next reproduce since I "sort of" know what to look for. Below is what I have on the most recent restore of services.
Any logging in particular you would need?
I feel like you're probably right about it being an issue with MetroNet. I called them today and they just rebooted everything and suggested I move to another piece of equipment that they can support.
Appreciate you linking me to that HOWTO post. I'll probably end up implementing that after I reproduce one more time.
This was on the latest restore:
(Post was flagged for SPAM, so created a pastebin)
https://pastebin.com/04r3wLek -
Yeah I bet it is that issue. The only way to be sure there is to packet capture the DHCP requests
and check that unicast packets are being ignored but broadcasts see responses. -
@stephenw10 The solution posted in that forum post - do you think a note can be added in the documentation?
"Turn gateway monitoring back on. Your issue is not with that. It's with Metronet DHCP relays not responding to unicast renewals, the logs just confirm my suspicions. Perform the following. Goto interfaces > WAN and under DHCP client configuration check the box "Advanced configuration" and under presets select FreeBSD default. Then further down under Lease requirements and requests in the box "Option Modifiers" enter the following supersede dhcp-server-identifier 255.255.255.255
-
Yes, something needs to be added there. At the very least it needs more eyes to assess whether or not what they are doing can ever be valid. And if it is what we could/should be doing to address that. Let me open a ticket....
Actually that dhclient option was added specifically to address this issue:
https://redmine.pfsense.org/issues/7416Mmm, not sure where I'd expect a note in the docs to be. In the config override section here maybe?
https://docs.netgate.com/pfsense/en/latest/interfaces/configure-ipv4.html#dhcp -
Believe you are 100% correct here. I didn't reproduce my symptom, but feel like manually restarting the interface proves what's happening.
In the packet captures, when source is my IP only DHCP Request packets get sent out with no DHCK ACK's (I saw these when I left packet capture going overnight). When source is 0.0.0.0 and destination is 255.255.255.255 I get DHCP ACK's back (saw this when manually restarting the WAN interface)
Your help is much appreciated.
-
Yes, that sounds very much like you're hitting that issue. Try setting the supersede option and see if it returns.
