[10:51] <antaresgades> Hello. I'm trying to install 20.04 on a old hp dc7800 machine. I have a issue relating using of ahrdware RAID. I defined that RAID1 with the BIOS Intel Matrix Storage Manager, but the installer is not able to see it when partitioning etc. Moreover, if I try to install the OS in one of the disks, the installation breaks (there are some complaints
[10:51] <antaresgades> about finding partitions)
[10:52] <antaresgades> I've found that I should ensure to use the ubuntu SERVER disks for installation and of course I'm sure that I'm using it
[10:52] <antaresgades> andy advice?
[10:53] <lordievader> Can you drop to a shell and see if the kernel is seeing any drives?
[10:57] <ducasse> which files do i need to copy across to move a libvirt/kvm vm from one host to another?
[11:00] <antaresgades> I've dropped and made a ls /dev and I see devices md, md126, md127, sda and sdb
[11:00] <antaresgades> and also with dmesg I see at least it see the drives
[11:01] <antaresgades> and explicitely, some lines refer to md/raid1_md126 active with 2 out of 2 mirrors and many lines more
[11:03] <antaresgades> with fdisk -l it shows /dev/sda, /dev/sdb and /dev/md126
[11:05] <lordievader> ducasse: Depends on your setup.
[11:05] <lordievader> antaresgades: And you don't see those in the server installer?
[11:05] <lordievader> ducasse: Are you using qcow based files or lvm volumes or something for your vms?
[11:06] <ducasse> lordievader: qcow
[11:08] <antaresgades> lordievader: That's the issue. I'm offered the options USE an ENTIRE DISK in "guided storage configuration" and no RAID is present. And I I choose "Custom storage layout" , the two physical devices are present but no RAID.
[11:08] <lordievader> ducasse: In that case you want to transfer the qcow file and the xml definition of the VM. Import the xml definition on the other end. Adjust the config to reflect the new path to the qcow file.
[11:08] <antaresgades> lordievader: by the way, the installer version is 21.01.2
[11:09] <lordievader> antaresgades: What I remember from my last Ubuntu server install (mind you years back) was that there was no support for md-raid/lvm in the 'normal' image, but there was in the alternate server image.
[11:09] <ducasse> lordievader: where are the xml files stored by default?
[11:10] <lordievader> It could be that there, still, is no support for md/lvm in the normal image. (Don't ask why not, beats me)
[11:10] <antaresgades> I will try then. But maybe the right question here is: Would I take any advantage from BIOS RAID over software RAID? Do you have any experience on this? Maybe it is not worth the pain
[11:12] <lordievader> ducasse: `/var/lib/libvirt` probably, but you rather want to use the `virsh dumpxml` command.
[11:12] <lordievader> antaresgades: Hardware raid is faster (but has downsides). Software raid is more flexible (especially if you use LVM).
[11:12] <ducasse> k, thanks
[11:13] <ducasse> do i use it to export the qcow too?
[11:14] <lordievader> No, that is just to export the definition/config of the VM. Not the disk itself.
[11:18] <ducasse> right, got it
[11:21] <ducasse> lordievader: do i import the xml with virsh too?
[11:24] <lordievader> Yes
[11:25] <antaresgades> lordievader: Taking into account that I'm in a hurry with this... I'll dive into software RAID and see what happens;)  However, I've read that a recommended setup is having a software RAID and LVM on top. Is it possible to achieve with two diffffrente ccapacity (250 vs 500 gb) disks?
[11:26] <lordievader> Wait a minute, you said md drives right? That sounds like software raid already
[11:42] <RoyK> antaresgades: uses mdadm to create raids and then put lvm on top and then filesystems on lv's. mixing disks with different sizes is a bit trickyer, since a raidset can only consist of similar disk sizes. but if you must, you can partition the larger drives and use those partitions with the smaller drives and then add another raid with the leftover partitions. then use lvm on top to merge it all
[11:43] <RoyK> together. it's a bit messy, but it works
[11:45] <antaresgades> lordievader: for what I've reading, it looks like BIOS RAID are more or less like a software at the end. Maybe tht's why is seen tht way by the installer
[11:46] <antaresgades> RoyK: Yes it looks like a little mess. However, I've found here a guide for creating RAID with the installer with 20.04. At least I'm able to begin to the installation, see what happens
[11:46] <antaresgades> https://askubuntu.com/questions/1234949/install-ubuntu-20-04-focal-fossa-with-raid-1-on-two-devices
[11:47] <antaresgades> But is a mess... all this happens when you must assembly a frankenstein savenging old computers from one day to the next :DD
[11:52] <lordievader> antaresgades: I'd just use lvm, also for the raids. LVM is using md-raid for raids anyways.
[11:52] <lordievader> LVM allows you to specify per logical volume if you want it in raid, and what kind of raid.
[11:53] <antaresgades> I'll give a try then
[13:21] <RoyK> oh - he left
[16:47] <xnox> cpaelzer:  i'm trying to figure out how to set Ubuntu circle of friend in https://uefi.org/specs/ACPI/6.4/05_ACPI_Software_Programming_Model/ACPI_Software_Programming_Model.html#boot-graphics-resource-table-bgrt
[16:47] <xnox> so far i don't see anything in qemu
[16:47] <xnox> and i think it is possible in EDK2 build. But I don't see a nice way of doing it, apart from like replacing the tianocore stock image.
[16:49] <xnox> any ideas on how to make edk2 build to use Ubuntu logo by default?
[18:46] <MIF> question, I just frozzed my self out of my server
[18:46] <MIF> is there a way to get into it?
[18:49] <Ussat> terminal ?
[18:49] <Ussat> you cant ssh in is what you mean ?
[18:49] <MIF> no I can't
[18:50] <Ussat> So...terminal it is. Is this a VM ?
[18:50] <MIF> No
[18:50] <MIF> I am going to go and restart it
[18:50] <Ussat> Gluck
[18:53] <Ussat> How, if you dont have access ?
[18:54] <MIF> I have the physical server
[18:54] <MIF> but I can't even loggin via keyboard
[20:29] <daftykins> hello folks, i'm trying to get cifs multiuser working today - so far everything works correctly if i SSH into the host as a user, then add the password manually using cifscreds - i'm following this: https://serverfault.com/questions/392619/how-to-map-windows-acl-to-linux-ones-on-a-cifs-share
[20:30] <daftykins> things aren't working once i reached step 7 however, which was tying this into PAM to have it automatically add the password to the keyring at login, logs show this - "request-key[40885]: : No matching action"
[20:38] <Ussat> MIF, what do you mean you cant log in via keyboard, what happens when you try ?
[20:46] <daftykins> /etc/request-key.d/ contains 2 cifs related files provided by the package cifs-utils ; understanding why this doesn't work out of the box is beyond me :(
[20:52] <TJ-> daftykins: did you try the note "Before the second line, the existence of a permanent session keyring should be ensured ..." ?
[20:54] <daftykins> TJ-: both /etc/pam.d/login and /etc/pam.d/sshd contain this line so i assumed it'd be ok "session    optional   pam_keyinit.so force revoke"
[20:55] <TJ-> daftykins: is it failing for ssh and console logins?
[20:56] <daftykins> TJ-: as it's a remote server, the SSH case is the only one relevant to me really - i do see the error i posted twice immediately, but on a user logging in i do see a positive looking line in logs: sshd[40778]: pam_cifscreds(sshd:session): credential key for \\192.168.10.3\username added
[21:00] <TJ-> daftykins: that's what I meant - if one works and one doesn't that points to a problem in the failing access (SSH) rather than a failure in your overall configuration
[21:00] <TJ-> daftykins: try forcing SSH to use 'required'  "session required pam_keyinit.so"
[21:02] <daftykins> ok testing
[21:03] <daftykins> TJ-: nah, identical output
[21:04] <TJ-> daftykins: see "man 8 pam_keyinit" -- try adding the 'debug' option maybe?
[21:04] <daftykins> yeah debug is on in order to get the "No matching action" output
[21:05] <daftykins> either that or i'm checking for logs in the wrong way, right now i'm just reading the end of "journalctl -xe"
[21:05] <TJ-> has the system any other non-default PAM settings added?
[21:05] <daftykins> nah, vanilla system otherwise
[21:06] <TJ-> I'm wondering if there's another PAM module interferring due to ordering
[21:06] <daftykins> unless R-Studio has done anything in order to tie its' logins to it
[21:38] <daftykins> TJ-: for fun i created a debian 10 VM and that behaves identically
[21:39] <TJ-> fails for ssh but not for console?
[21:40] <daftykins> TJ-: no, fails on both
[21:44] <TJ-> hmmm, would something like this (with correct UID) give clues? " journalctl -u user@1000.service "
[21:45] <daftykins> i sub'd that to be my username and left it as 1000 since that's my UID, do i need to change .service to something else as well? i just got no output at all
[21:49] <daftykins> oh my deary me i solved it
[21:50] <daftykins> let me test again to confirm
[21:53] <daftykins> ok it works on my debian VM logging in at a TTY, but not over SSH
[21:56] <daftykins> that's really weird
[22:01] <TJ-> daftykins: right, so it's the unique path PAM takes for SSH vs login (console) - compare the options and figure out what ssh is missing!
[22:02] <daftykins> btw my mistake before was that i was specifying domain=<server IP> in the pam.d files but my .smbpass credential file didn't declare the domain, i added that param "domain=IP" and it began working
[22:03] <daftykins> TJ-: i've no idea how to do that part :D
[22:06] <daftykins> weeeeird now both work on my debian VM