| 1 |
On 12/11/2018 04:23 AM, Tsukasa Mcp_Reznor wrote: |
| 2 |
> You're totally correct, more information would be beneficial, here goes. |
| 3 |
|
| 4 |
:-) |
| 5 |
|
| 6 |
> All machines are Wired 1Gbps connections. |
| 7 |
|
| 8 |
ACK |
| 9 |
|
| 10 |
That means that you don't have the complications (and performance |
| 11 |
issues) of wireless. |
| 12 |
|
| 13 |
> Uefi IP4 network stack sends dhcp request, gets boot file pxelinux.efi, |
| 14 |
> the default entry sends the linux kernel (no initramfs needed, firmware |
| 15 |
> added to kernel image). |
| 16 |
|
| 17 |
Interesting. |
| 18 |
|
| 19 |
Do you have reservations in the DHCP server? Or are the addresses truly |
| 20 |
dynamic? |
| 21 |
|
| 22 |
Are you relying on the client's UEFI implementation to provide the menu? |
| 23 |
Or are you using PXELINUX for the menu? (I know it's a nuance, but it |
| 24 |
is a difference.) The latter is much easier to centrally manage than |
| 25 |
the former. |
| 26 |
|
| 27 |
Does the UEFI stack get the same IP via DHCP that the OS gets via DHCP? |
| 28 |
Is there any sort of contention? Does UEFI release the IP before |
| 29 |
bootstrapping the PXELINUX image? Does the DHCP server view the |
| 30 |
multiple requests from the same client MAC as a form of a refresh? Or |
| 31 |
does it just offer the same IP? |
| 32 |
|
| 33 |
> Another good note is the kernel contains the command line built-in for |
| 34 |
> using root on NFS. |
| 35 |
|
| 36 |
Okay. ~pondering~ |
| 37 |
|
| 38 |
Are all clients booting the same kernel, thus using the same command line? |
| 39 |
|
| 40 |
This means that clients must use DHCP to retrieve their IP address. |
| 41 |
|
| 42 |
I guess there is some opportunity to return different files (PXELINUX |
| 43 |
image / config and / or kernel file) to different clients to get |
| 44 |
different behavior. But that might be more complexity than is necessary. |
| 45 |
|
| 46 |
Would you please share the kernel command line? I'm quite curious what |
| 47 |
the syntax is for NFS root. |
| 48 |
|
| 49 |
> Machine loads, mounts the required mount points through NFS4.2 (so much |
| 50 |
> better than the old NFS 3 speeds). |
| 51 |
|
| 52 |
Nice. |
| 53 |
|
| 54 |
> LightDM loads and users are free to work, in this case family members |
| 55 |
> playing Steam/Diablo 3/etc. |
| 56 |
|
| 57 |
:-) |
| 58 |
|
| 59 |
> I switched to using Root on NFS for alot of reasons. |
| 60 |
|
| 61 |
:-) |
| 62 |
|
| 63 |
> Maintaining 4 gentoo installs on machines of varying specs and remembering |
| 64 |
> to update each with good updates added a fair amount of administration |
| 65 |
> time. (4, because the server is included) |
| 66 |
|
| 67 |
*nod* |
| 68 |
|
| 69 |
> Using chroots on the server as binary build hosts for each machine |
| 70 |
> solves some problems, but increases space requirements quite a bit, and |
| 71 |
> adds latency if you want to use it while it's emerging anything, plus |
| 72 |
> compiling say Libreoffice or whatever 3+ times in a row is pretty slow. |
| 73 |
|
| 74 |
That makes me think that you are using a separate NFS export for each |
| 75 |
machine's root. |
| 76 |
|
| 77 |
I have wondered about trying to do something similar (likely start in a |
| 78 |
VM) that has (at least) one machine specific export for things like |
| 79 |
/etc, but would then try to use a common export for things like /usr, |
| 80 |
/lib, and maybe /var. |
| 81 |
|
| 82 |
Maybe a common / export and a per machine /etc would accomplish what I'm |
| 83 |
thinking. |
| 84 |
|
| 85 |
> Side note, If anyone else runs diskless I have a patch for wine I can |
| 86 |
> send out that returns the nfs mount as a fixed hard drive, there are a |
| 87 |
> few apps/games that refuse to install/run on a network share, and a patch |
| 88 |
> for steam that removes the file locking issues so updates run quick and |
| 89 |
> smooth |
| 90 |
|
| 91 |
Nice. |
| 92 |
|
| 93 |
> (neither will ever be upstreamable, people have tried in the past) |
| 94 |
|
| 95 |
:-/ |
| 96 |
|
| 97 |
> Thanks for your response, I'd love to help if you have any more questions, |
| 98 |
> it's been a fun experience for me for sure. Also, cachefilesd if there's a |
| 99 |
> drive available, makes everything feel like it's not a networked machine |
| 100 |
> at all here. |
| 101 |
You're welcome. |
| 102 |
|
| 103 |
Thank you for sharing. |
| 104 |
|
| 105 |
I'd love to know more about how you're doing things. |
| 106 |
|
| 107 |
- What is common between the diskless clients and what is unique. |
| 108 |
- PXELINUX image / config |
| 109 |
- Kernel image |
| 110 |
- NFS exports |
| 111 |
- What do your exports look like. |
| 112 |
- What sort of configuration you have in your DHCP server that's |
| 113 |
specific to this. |
| 114 |
- Any sticky reservations, possibly with machine specific parameters. |
| 115 |
- Other things that I can't think of at the moment. |
| 116 |
|
| 117 |
Thank you again. Very interesting stuff. |
| 118 |
|
| 119 |
|
| 120 |
|
| 121 |
-- |
| 122 |
Grant. . . . |
| 123 |
unix || die |
Archives