IRC chat logs for #ltsp on irc.libera.chat (webchat)
Channel log from 16 June 2023 (all times are UTC)
| 04:55 | <vsuojanen[m]> Good morning, have you experienced desktop for ltsp client load slow during login with gdm3 and nfs home dir ? I have couple ThinkCentre M90n-1, attached to older monitor with DP-VGA adapter. the issue is especially when only when I boot with nfs home dir, local boot works better. two clients have slightly slow network connections 1Gbps but one has good (1Gbps i believe)
| |
| 04:58 | <quinox> My load times increased considerably when I switched the server to SSDs
| |
| 05:02 | <vsuojanen[m]> i have configured AUTOLOGIN and nfs home looks mount fine every time, desktop session exec starts fine but graphics don't load, just black screen with mouse pointer, when i reboot the client it loads the graphics like ok like with any other client hardware. issue is with this ThinkCentre M90n-1
| |
| 05:07 | and it's first network boot e.x. first network boot after I have booted first the client with the local OS and then rebooted from network
| |
| 05:13 | <quinox> If you have have jumbo frames enabled: not all clients support it, and when they don't they suffer
| |
| 05:14 | wyre is back | |
| 05:15 | <quinox> although that doesn't jive with the NFS part of your story, I think that runs on TCP
| |
| 05:18 | <vsuojanen[m]> so if the clients suffer from network that nfs part would run slow?
| |
| 05:20 | like if the issue was caused with the jumbo frames enabled
| |
| 05:25 | 2: enp2s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT group default qlen 1000
| |
| 05:25 | <alkisg> vsuojanen: do this and tell us the result: https://epoptes.org/documentation/lan-benchmark/
| |
| 05:29 | * vsuojanen[m] uploaded an image: (65KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/GupPmBILahBcbeEfFnZeOhGA/image.png > | |
| 05:29 | <vsuojanen[m]> 2: enp2s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT group default qlen 1000
| |
| 05:29 | 3: enp4s0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc fq_codel state DOWN mode DEFAULT group default qlen 1000
| |
| 05:33 | it's physically connected in same switch with the ltsp server, which is lxc container bridged to the host bridge with bonded interface which has two phy nic's bridged
| |
| 05:36 | * vsuojanen[m] uploaded an image: (53KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/yVMfeGSgMhiclwaDgoqABrkN/image.png > | |
| 05:36 | <alkisg> vsuojanen: do two (or even better four) of them at the same time
| |
| 05:36 | Otherwise you won't be able to see flow control issues
| |
| 05:36 | <vsuojanen[m]> e.x. other client with different hardware in same switch boots with speeds like this ^
| |
| 05:40 | should I be able to launch two of them in epoptes same time?
| |
| 05:41 | never mind I figured to select multiple clients first
| |
| 05:42 | <alkisg> Right
| |
| 05:46 | wyre is now away: Auto away at Fri Jun 16 05:45:19 2023 UTC | |
| 05:47 | * vsuojanen[m] uploaded an image: (148KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/coAQJcaLRULbILOEMwdjjJSH/image.png > | |
| 05:47 | woernie has joined IRC (woernie!~werner@pd9e8b2a5.dip0.t-ipconnect.de) | |
| 05:48 | <vsuojanen[m]> two clients in same switch, that second slow upload is the thinkcentre model with issues
| |
| 05:48 | *four clients
| |
| 05:48 | <alkisg> Try to move it to another network socket
| |
| 05:49 | If it persists being slow after that, then it's a NIC driver issue
| |
| 05:50 | <vsuojanen[m]> thanks for help. I was about to do some work there actually, I will be back in 1-2 hours
| |
| 05:55 | ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz) | |
| 07:26 | wyre is back | |
| 08:22 | wyre is now away: Auto away at Fri Jun 16 08:20:40 2023 UTC | |
| 08:25 | wyre is back | |
| 08:40 | <vsuojanen[m]> alkisg:thinkcentre model client boot still had similar issues in another port on switch. we settle with it and look some time results from distro and firmware updates.
| |
| 08:40 | * vsuojanen[m] uploaded an image: (71KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/EJOmtycujfEqRAjNEqzNpHlL/image.png > | |
| 08:40 | <alkisg> When you tested with a local installation, where you using an SSD disk?
| |
| 08:41 | Also are you using round-robin or xor, on your server bond?
| |
| 08:41 | Or 802.3ad?
| |
| 08:43 | <vsuojanen[m]> thinkcentre m90-n1 it's internal nvme storage. the server bond is balance-tlb(5)
| |
| 08:44 | <alkisg> OK nvme has 2500 mbps while your client gets 230 mbps; nvme will be faster :D
| |
| 08:45 | But do try with the xor bonding method, I think it may solve the 42 mbps upload
| |
| 08:47 | <vsuojanen[m]> thanks let's see or 802.3ad
| |
| 08:51 | <adamjurkiewiczpy> vsuojanen: and write how its ends.,. It is always good to know solutions
| |
| 08:56 | <vsuojanen[m]> in general it is good practice but I expect this is some sort of client hardware firmware issue.
| |
| 09:20 | <adamjurkiewiczpy> Then write about it.
| |
| 09:22 | wyre is now away: Auto away at Fri Jun 16 09:20:33 2023 UTC | |
| 09:29 | wyre is back | |
| 10:14 | wyre is now away: Auto away at Fri Jun 16 10:13:08 2023 UTC | |
| 11:22 | wyre is back | |
| 11:43 | wyre is now away: Auto away at Fri Jun 16 11:41:39 2023 UTC | |
| 15:08 | wyre is back | |
| 15:39 | wyre is now away: Auto away at Fri Jun 16 15:37:56 2023 UTC | |
| 16:57 | vagrantc has joined IRC (vagrantc!~vagrant@2600:3c01:e000:21:7:77:0:50) | |
| 18:45 | wyre is back | |
| 19:23 | wyre is now away: Auto away at Fri Jun 16 19:22:21 2023 UTC | |
| 20:59 | ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving) | |
| 22:36 | Vercas has left IRC (Vercas!~Vercas@gateway/tor-sasl/vercas, Remote host closed the connection) | |
| 22:36 | Vercas has joined IRC (Vercas!~Vercas@gateway/tor-sasl/vercas) | |