00:15 | vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Quit: leaving) | |
05:40 | ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz) | |
05:42 | RaphGro has joined IRC (RaphGro!~raphgro@fedora/raphgro) | |
07:17 | woernie has left IRC (woernie!~werner@p5ddec80a.dip0.t-ipconnect.de, Ping timeout: 260 seconds) | |
07:18 | woernie has joined IRC (woernie!~werner@p578bb7b6.dip0.t-ipconnect.de) | |
07:48 | alkisg has left IRC (alkisg!~alkisg@ubuntu/member/alkisg, Remote host closed the connection) | |
07:51 | alkisg has joined IRC (alkisg!~alkisg@ubuntu/member/alkisg) | |
07:57 | woernie has left IRC (woernie!~werner@p578bb7b6.dip0.t-ipconnect.de, Ping timeout: 246 seconds) | |
07:58 | woernie has joined IRC (woernie!~werner@p5ddec80a.dip0.t-ipconnect.de) | |
08:37 | alkis[m] has left IRC (alkis[m]!alkismatri@gateway/shell/matrix.org/x-nncvbvjkcuawegzu, Quit: authenticating) | |
08:37 | alkis[m] has joined IRC (alkis[m]!alkismatri@gateway/shell/matrix.org/x-nmzjmeultgjbufgd) | |
11:36 | adrianorg has left IRC (adrianorg!~adrianorg@179.187.31.189, Ping timeout: 252 seconds) | |
11:37 | adrianorg has joined IRC (adrianorg!~adrianorg@179.187.28.136) | |
12:00 | woernie has left IRC (woernie!~werner@p5ddec80a.dip0.t-ipconnect.de, Remote host closed the connection) | |
12:30 | RaphGro has left IRC (RaphGro!~raphgro@fedora/raphgro, Quit: Please remember your own message. It'll be read as soon as possible.) | |
13:12 | woernie has joined IRC (woernie!~werner@p200300c18f2d717755b4128a3073d70a.dip0.t-ipconnect.de) | |
13:59 | Monkberry has joined IRC (Monkberry!6c221083@static-108-34-16-131.bflony.fios.verizon.net) | |
14:01 | <Monkberry> hey all
| |
14:02 | <alkisg> Hi Monkberry
| |
14:03 | <Monkberry> NFSD: client 172.16.11.184 testing state ID with incorrect client ID
| |
14:04 | Has anyone experienced this?
| |
14:04 | It seems to cause all sorts of issues on the network
| |
14:04 | <alkisg> Monkberry: I think I've seen it when a client was forcibly shut down, without notifying nfs, then another one booted with that same IP
| |
14:05 | So the server was thinking "hey, wasn't there another one using ip=.184?"
| |
14:05 | <Monkberry> hmm, that wouldn't surprise me there
| |
14:05 | <alkisg> Make sure that your DHCP server is respecting IP leases etc
| |
14:06 | <Monkberry> hmmm, might this have something to do with what we did to work around the vlans?
| |
14:06 | * alkisg tries to remember the workaround... | |
14:06 | <Monkberry> Most seem to come from that side of the fence
| |
14:06 | adrianor1 has joined IRC (adrianor1!~adrianorg@187.113.216.137) | |
14:07 | adrianorg has left IRC (adrianorg!~adrianorg@179.187.28.136, Ping timeout: 240 seconds) | |
14:07 | <Monkberry> One side (the side seemingly with less troubles has dhcp from a pfsense box
| |
14:07 | The other side of the vlan has dhcp handed out from a server running isc-dhcp-server
| |
14:08 | <alkisg> And the ranges are different of course, right?
| |
14:09 | I think we only changed the reported server; we didn't touch anything about ip leases
| |
14:09 | <Monkberry> yes. I have the changes we made
| |
14:09 | <alkisg> Monkberry: can we vnc? It'll be faster...
| |
14:09 | x11vnc -connect alkisg.ltsp.org
| |
14:12 | <Monkberry> That box is not the ltsp server
| |
14:12 | that one is on the other vlan where dhcp is from isc-dhcp-server
| |
14:14 | <alkisg> Monkberry: this is where the most failures happen?
| |
14:14 | <Monkberry> no, the other server on the other vlan
| |
14:14 | I can get you connected to that
| |
14:14 | <alkisg> Sure, let's
| |
14:17 | Monkberry: and there, pfsense assigns the IPs? Are they hardcoded, why doesn't it answer now?
| |
14:17 | <Monkberry> Yes, pfsense hands out on that side of the vlan
| |
14:17 | I'm not sure what you mean, hardcoded and not answering
| |
14:18 | <alkisg> I tell the ltsp server to do a fake dhcp request
| |
14:18 | And pfsense doesn't send an IP
| |
14:18 | <Monkberry> idk
| |
14:18 | <alkisg> ok
| |
14:18 | <Monkberry> I can get you logged into pfsense so you could see that
| |
14:19 | <alkisg> Can I reach a client on the problematic vlan via epoptes?
| |
14:19 | <Monkberry> yes
| |
14:19 | one sec
| |
14:19 | <alkisg> ok
| |
14:21 | valid_lft 4277sec preferred_lft 4277sec
| |
14:23 | Monkberry: I think you have a very very small lease time
| |
14:23 | And then the server doesn't even hand out the same IP
| |
14:23 | Let's connect to pfsense
| |
14:23 | <Monkberry> ok
| |
14:24 | <alkisg> Maybe you have e.g. 2 hours; it would be best to be at least a day; better yet, a week, so that clients that boot once per week get the same IP
| |
14:24 | <Monkberry> It's a very busy school and with everyone's phones, watches, pcs, I tweaked the lease time
| |
14:24 | <alkisg> Hmmm
| |
14:25 | Monkberry: let's go to pfsense
| |
14:25 | <Monkberry> I seem to have lost control of my mouse
| |
14:25 | <alkisg> Yeah me too I think we were both trying to use it at the same time
| |
14:25 | OK, watching...
| |
14:26 | Go to lease time, I don't recall where that is
| |
14:28 | Monkberry: so, ltsp clients can't really renew their leases as if they change their IP all the nfs etc connections will expire and they won't have access to /, so they won't be able to load programs etc, they'll just hang,
| |
14:28 | <Monkberry> I tweaked the lease time because I was running out of ip addresses and this was the easy fix for the class c that was in place
| |
14:28 | suggestion?
| |
14:29 | <alkisg> The current lease time is 1 hour. The initial was 7200, 2 hours?
| |
14:29 | And you had issues with lease time=2 hours?
| |
14:29 | If so, this means you need to grow your network, it's too stressed
| |
14:29 | <Monkberry> I'm thinking it may be possible to have the clients that get assigned addresses on a longer lease than those just getting from dhcp
| |
14:29 | <alkisg> Bite the bullet and increase the subnet :)
| |
14:30 | The ltsp clients may also have static mappings, one for each one of them
| |
14:30 | E.g. there's no point in having a pool of 20, when you have 30 clients. You should have 30 IPs for your 30 clients etc
| |
14:30 | (excluding the roaming/mobile ones)
| |
14:30 | <Monkberry> I'll look into the lease times in the meantime, I'll figure the summertime to increase the subnet, with this being a school
| |
14:31 | <alkisg> It shouldn't be intrusive though, if you just grow the existing subnet
| |
14:31 | <Monkberry> So, your determination is that error is being caused by dhcp lease times?
| |
14:31 | <alkisg> Yes, I think that pfsense gives the IP to another client
| |
14:31 | <Monkberry> How can I just grow the subnet without changing everything?
| |
14:32 | Is that even possible?
| |
14:33 | <alkisg> The 172.16.11.x/24 subnet would become e.g. 172.16.y.x/22
| |
14:34 | <Monkberry> That's what I thought
| |
14:34 | <alkisg> So you'd keep the same IPs everywhere; DHCP clients would get a different netmask; and if you have printers etc, then you'd need to update their netmask, otherwise they wouldn't be able to reply to the clients over the /24
| |
14:35 | <Monkberry> ah
| |
14:36 | Thank you very much for the help, I'll send you like I do
| |
14:37 | <alkisg> Monkberry: another idea is to put the ltsp clients with infinite lease time, let me see about that...
| |
14:37 | woernie has left IRC (woernie!~werner@p200300c18f2d717755b4128a3073d70a.dip0.t-ipconnect.de, Remote host closed the connection) | |
14:38 | <Monkberry> Some of those are not on the ltsp server, however, they are sharing home directories
| |
14:38 | <alkisg> inet 10.32.4.229/21 brd 10.32.7.255 scope global enp1s0
| |
14:38 | valid_lft forever preferred_lft forever
| |
14:39 | <Monkberry> Like the room 104 that you were looking at. They are a stand alone Linux box (Mint 20) with a home directory that is mounted from the server
| |
14:39 | <alkisg> This is what a client lease looks like, in another installation with pfsense that I maintain,
| |
14:39 | which means that ltsp clients there never need to renew their leases; i.e. it's possible to configure that in pfsense
| |
14:39 | So, do you have enough IPs for all the school PCs to be on at the same time?
| |
14:39 | If yes, you can assign them infinite leases
| |
14:40 | (essentially, static IPs but managed in pfsense)
| |
14:40 | <Monkberry> That depends on the kids coming in with their watches and tablets now
| |
14:40 | <alkisg> You wouldn't assign THESE an infinite leases
| |
14:40 | Only to your known school PCs
| |
14:41 | <Monkberry> Yes I know but on some days I think they just run out of ips
| |
14:41 | <fiesh> just jumping in now, it seems weird to me "official" computers are assigned in the same lease range as random bring-your-own-stuff devices...
| |
14:41 | <Monkberry> I'll most likely need to increase the class
| |
14:41 | <alkisg> fiesh: Monkberry only has 255 IPs there, he'll update that in the summer
| |
14:42 | Monkberry: since you already had issues with ip lease time = 2 hours, I'd really look into increasing the subnet immediately. Also, removing the vlans :D
| |
14:42 | <Monkberry> I'd love to remove those vlans
| |
14:42 | <alkisg> You'd need one hour when the school is closed, but it should suffice
| |
14:42 | <fiesh> ok I have no idea what's going on, but if I had trouble with having sufficiently many IP addresses and couldn't increase the subnet for some reason and wanted to give random people access, I'd set up an additional NAT for their wifi
| |
14:43 | and give my important devices proper addresses
| |
14:43 | <alkisg> Right ^
| |
14:43 | That's a good idea as well
| |
14:43 | In some schools, we have a different access point and subnet for roaming clients
| |
14:43 | <fiesh> like this, someone could just come and grab all leases, and your *official* equipment would stop working
| |
14:43 | <Monkberry> Yes that is a good idea flesh, separate the wifi
| |
14:43 | <alkisg> E.g. school is 10.x.y, while roaming= 192.168.1.x
| |
14:44 | Many access points support that without even involving pfsense
| |
14:44 | <fiesh> you have an actual /24 network you're using?
| |
14:44 | <Monkberry> They have a bunch of ubiquity unifi devices throughout the school that hand off the wifi
| |
14:44 | yes
| |
14:45 | <fiesh> well then you're loaded with IP addresses anyway by my standard ;-)
| |
14:45 | <Monkberry> 2 - /24 on 2 - vlans
| |
14:45 | <fiesh> you have two actualy /24 blocks, wow... I'd never give out actual addresses to random wifi users
| |
14:46 | but then I only have 8 addresses ;-)
| |
14:46 | but I know that back in the US, every printer had its own address since the distribution was so generous amongst educational institutions hehe
| |
14:47 | <Monkberry> They used to have printers throughout the hallways when I came in
| |
14:47 | windows pcs that they would never log out of, a complete hodge podge of crap
| |
14:49 | It's really quite a place now. A school in the US that is completely running Linux on pcs, voip phones, cameras, etc..
| |
14:49 | <fiesh> well still I'd never give out actual IPv4 addresses to random wifi people -- IPv6 is fine I guess
| |
14:50 | <Monkberry> Anyway, thank you all for the help. I gotta run but this is food for fodder in the right direction.
| |
14:50 | <fiesh> really great though, wish there were laws that required public money to only be spendable on open source software
| |
14:50 | :) bye
| |
14:51 | <Monkberry> I could tell you many stories of all the fights I've had over this but they believed me and now they love it
| |
14:51 | See you later and thanks again!
| |
14:51 | Monkberry has left IRC (Monkberry!6c221083@static-108-34-16-131.bflony.fios.verizon.net, Quit: Connection closed) | |
20:54 | Helenah has left IRC (Helenah!~irc@unaffiliated/iveeee, Quit: Switching from weechat relay to ZNC) | |
22:12 | ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving) | |