17:00:29 #startmeeting fedora-server 17:00:29 Meeting started Wed Sep 15 17:00:29 2021 UTC. 17:00:29 This meeting is logged and archived in a public location. 17:00:29 The chair is pboyHB. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:29 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:00:29 The meeting name has been set to 'fedora-server' 17:00:39 #topic Welcome / roll call 17:00:48 Welcome to our IRC meeting! 17:00:57 As usual, we'll give a few minutes for folks to show up 17:01:05 Please, everybody who is lurking, say either .hello2 or .hello 17:01:14 .hello2 17:01:14 I’ll post the agenda in a few minutes. 17:01:15 jwhimpel: jwhimpel 'John Himpel' 17:02:17 jwhimpel: Welocme! 17:04:15 #topic Agenda 17:04:23 #link https://lists.fedoraproject.org/archives/list/server@lists.fedoraproject.org/thread/A5FMNBDH5R4ADXGZYVNV7TOGIBDP3FDJ/ 17:04:30 1. Follow up actions 17:04:38 2. Max size arm-32 exceeded, install media blocked (continuation) 17:04:47 3. Facilitated deployment of key services by combining rpm and Ansible 17:04:54 4. Open Floor 17:05:02 Any additional topic ? 17:05:55 OK. No additional topic 17:06:04 #topic Follow up actions 17:06:32 Only open action as of today: libvirt test day 17:06:46 Unfortunately, Langdon who took this into his hands, is busy today and can't participate. 17:07:00 Does anyone else got information about this? 17:08:25 Obviously no information available. 17:08:38 changing topic 17:08:45 #topic Max size arm-32 exceeded, install media blocked 17:08:54 #link https://bugzilla.redhat.com/show_bug.cgi?id=1963007 17:09:05 Latest status: there are 2 kernels included, one of the is PAE version which might be unnecessary for 32 bit ARM 17:09:17 We want to check kernel before simply increase image size 17:09:17 I posted a query on the ARM list. No one took the bait. 17:09:43 jwhimpel: Yes I read the list as well 17:10:18 Did someone else check the iso generation? 17:10:34 My very limited knowledge thinks one kernel is for very small SOC systems and the other kernel is for larger productional-capable ARM server boards. 17:11:42 jwhimpel: Yes. One question was, doe such production grade > 4gb systems exist for ARM 32? 17:11:59 and using Fedora? 17:12:05 I have no idea. 17:13:23 If nobody objects, we can only just increase the size limit and don't do any quality checks. 17:14:07 There seems to be several "nuts and bolts" ARM folks that follow on IRC. Perhaps, I could raise the kernel issue there. 17:14:41 jwhimpel: I would appreciate that. 17:15:33 Cant we give a week to try to get solid information and otherwise just increase the size limit? 17:15:56 I think, we can not wait undefinitely 17:15:58 Sounds reasonable to me. 17:16:57 #agreed We take a week to get additional information and otherwise just increase size limit. 17:17:46 #action jwhimpel contacts ARM folks again about possible ARM32 kernel issue 17:18:15 #topic Facilitated deployment of key services by combining rpm and Ansible 17:18:25 In particular, to discuss: Where and how we want to make our Ansible artifacts available. 17:18:34 Floor is open. 17:19:16 I am working to refactor the ansilble role to install and configure a basic wildfly instance. It's a slow slog, but I am making progress. 17:19:48 jwhimpel: Good to hear. (ehm to read) :-) 17:20:03 I still need to research ansible collections and how this effort might be distributed via ansible-galaxy. 17:21:09 I suppose, as soon as we have somethint to play with, the interest about where to make it available will raise considerable 17:21:44 My wish is, to keep a tight connection to Fedora Server edition. 17:23:00 It was suggested at the last meeting that it also be usable on RHEL. 17:23:32 That's a good idea, I think. 17:23:44 As long as we don't have to support RHEL6, the same ansible role should work for both without too much extra effort. 17:24:04 Isn't RH 6 EOL? 17:24:41 RHEL 6 has already passed several EOL milestones, but there is still some very limited support. 17:25:19 pboyHB: RHEL 6 is currently in ELS (extended lifecycle) 17:25:22 OK. But I think we can ignore RH 6. :-) 17:25:30 I read somewhere that RH is building or has built a large ansible collection. 17:26:06 I may try to support those versions they support there 17:26:14 It has. But from my limited browsing, it's rather basic. 17:26:41 Well, I didn't even browse. 17:27:38 But I suppose, RH should already have perfect ansible support for jBoss in place? 17:29:03 There was a recent article from RH on installing JBoss, but it involved interfacing with their subscription manager and other non-Fedora related issues. 17:30:59 Good, i think we can savely ignore that 17:31:12 Well, today nobody else seems to be ready to contribute the this topic. 17:31:27 If nobody objects I switch to text topic. 17:33:08 #topic 4. Open Floor 17:33:20 No infos from me here. 17:33:29 The floor is open. 17:35:07 rsc: Thanks. Recently I tried to get a version 6 from CentOS and Scientific Linux. No luck 17:36:09 I'm interested in hearing feedback on my email to the list about the documentation. 17:37:05 mowest: I suppose I overlooked / forgot something. Mea culpa! Can you give some hint? 17:38:43 It is titled "Feedback on Documentation After a Fedora Server Install" 17:40:35 Yes, I remember! You are Stephan Daley, right? 17:40:41 Yes 17:40:55 I wrote it a few days ago 17:41:53 Right not, BTTFS is not recommended for Fedora Server. There was a long discussion on the development email list about a month ago with pros and cons. It was all above my pay grade to understand. 17:42:02 Yes, 2 days ago 17:42:03 s/not/now/ 17:42:15 I didn't forget, but just managed to answer yet. A big sorry, I'm a bit swamped ( a lot of bit). It's on my todo tomorrow. 17:43:59 jwhimpel: agreed about BTRFS as the general file system. But probably for specific purposes. I use it for system-nspawn containers which make use of special BTRFS functions. 17:44:52 No worries, Peter, don't feel bad that you didn't get to my email yet. 17:45:16 mowest: thanks. 17:45:20 By the way, I plan to have a more detailed discussion about the progress of our documentation next meeting. 17:45:43 The section on "logging in via ssh to root", is probably a bit controversial. Some believe logging in a root is fine. Others believe it to be a huge security risk. Reminds me of the vi/emacs wars. 17:46:42 Yes, I think it depends on how many admins have access. 17:47:23 jwhimpel, didn't know that about Btrfs, is that up for continued discussion in the future for server? I believe that Btrfs is a great option to be pushing for the homelab people that are looking at Fedora Server as a possible more feature rich replacement for a Synology (that uses Btrfs) or TrueNAS (uses ZFS) box 17:48:18 jwhimpel: If you have 10 or more admins, loggin in as root may be a bad idea. Ig you have only 2 or if you are the only one (as in a home lab server) 17:48:47 jwhimpel, totally agree with the not logging in as root. I didn't do that myself. I logged with with a user account that had sudo privileges and this exchanged keys using ssh-copy-id which is what most tutorials seems to encourage. 17:49:17 Regarding BTRFS: The professional server people are quit conservative regarding file systems. For them security and reliablility is top! 17:49:25 It was not a final decision. Rather, I believe it was a wait and see what (if any) issues the desktop folks experience since they are implementing this first. Server folks are a conservative bunch. 17:49:34 And BTRFS has still some issues there. 17:49:53 And has some issues as a file system for databases 17:50:24 pboyHB: databases work fine with nocow (which is what one should use) 17:51:14 dcavalca: thanks, I didn't know that. Just have read a article written by a coreos engineer. 17:52:15 And as far as I know coreos is hestent to use BTRFS 17:52:17 Makes some sense to not have as default, perhaps it could be a separate section in the documentation if someone would like to use Btrfs, but I still feel like giving a little more guidance on a default partitioning scheme for those new to Fedora Server would be beneficial, because I felt as if I was jumping around in the docs a bit to try to figure 17:52:17 out a partitioning scheme that would work and give me a working RAID pool. 17:53:01 the main weakness in btrfs for Server is raid56, which is still experimental 17:53:06 but that's being actively worked on 17:53:26 Even if we had more detail for an LVM using xfs or ext4 as the file system, would be nice for someone new to Fedora Server. 17:53:32 raid 0/1/10 works fine (or one can do 5/6 with md on top of they so desire) 17:54:31 dcavalca, yes, I set up a RAID10 on Btrfs for my homelab experimental server. 17:54:44 Another topic is separation of concerns, here separation of system and user data in to completely independent file sysstems. 17:55:36 The goal is to prevent the system-wide spread of errors. 17:55:38 Yes, I mentioned that in my email too, because I wanted to have my system separate from my data. 17:56:22 fwiw, if you're using btrfs you can use subvolumes for that while keeping a single filesystem 17:56:45 of course, nothing stopping you from making multiple partitions/filesystems if you prefer 17:57:38 As far as I have read, subvolumes do not provide complete separation. 17:58:06 dcavalca: That's my solution for the time being. 17:58:51 dcavalca in this server, I'm using old drives that could fail. I wanted to have my system or / on one drive while having a Btrfs pool of 4 data drives so that if I had the / drive fail, I could just replace that drive with another, install Fedora Server and reconnect to the data pool. If one of the drives in the data pool failed, I could replace 17:58:52 that and let the RAID10 rebuild. This was my thinking, not sure if that is wise or not. 18:00:25 mowest: yeah, if you're using separate drives that seems fine, though it'd probably work just as well if you put / on the raid10 too and just made it a subvol 18:00:52 in that case, you'd want to add to fstab the knob to allow it to boot even with a degraded array (I forgot the name offhand) 18:01:38 Folks, time is up. It was a very interesting discussion in "open floor" today. Much appreciated. 18:01:49 Next meeting in 3 weeks!!! Otherwise we will lose our rhythm of 1st and 3rd Wednesdays. 18:02:21 If nobody objects, I'll close in2 minutes. 18:02:41 Anyway, I'm hoping that someone who understands partitioning and files systems would be willing to write a nice "Here is a disk partitioning scheme to get you started" using whatever sane defaults this working group has determined and perhaps LVM so that those new to using Fedora as a Homelab server would have a place to start. I just found that to 18:02:42 be overly challenging to figure out on my first install. 18:03:27 Thanks for the great work of everyone. Thanks for letting me participate in the open floor. Have a great two weeks. 18:03:33 mowest: I'll write my solutions in my answer. 18:03:51 mowest: You are very welcome! 18:03:58 Take your time, everyone is busy with real life too. :-) 18:04:20 By By everybody! 18:04:35 #endmeeting