Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Sep 13 2016 13:04

    galthaus on develop

    update metadata on test barclam… Merge pull request #1296 from z… (compare)

  • Sep 13 2016 13:04
    galthaus closed #1296
  • Sep 13 2016 04:52
    VictorLowther commented #1296
  • Sep 13 2016 04:52
    zehicle labeled #1296
  • Sep 13 2016 04:52
    zehicle opened #1296
  • Sep 13 2016 04:52
    zehicle assigned #1296
  • Sep 06 2016 20:50

    galthaus on develop

    :3000 no longer needed Merge pull request #1295 from z… (compare)

  • Sep 06 2016 20:50
    galthaus closed #1295
  • Sep 06 2016 20:48
    VictorLowther commented #1295
  • Sep 06 2016 20:48
    zehicle opened #1295
  • Sep 06 2016 06:11

    galthaus on develop

    allow prefix more control in no… Merge pull request #1294 from z… (compare)

  • Sep 06 2016 06:11
    galthaus closed #1294
  • Sep 06 2016 03:18
    VictorLowther commented #1294
  • Sep 06 2016 03:18
    zehicle opened #1294
  • Sep 01 2016 21:05

    galthaus on stop-dns-and-gssapi-in-sshd-in-sledgehammer

    (compare)

  • Sep 01 2016 21:05

    galthaus on develop

    Update sshd in Sledgehammer to … Merge pull request #1293 from d… (compare)

  • Sep 01 2016 21:05
    galthaus closed #1293
  • Sep 01 2016 21:03
    VictorLowther opened #1293
  • Sep 01 2016 21:03

    VictorLowther on stop-dns-and-gssapi-in-sshd-in-sledgehammer

    Update sshd in Sledgehammer to … (compare)

  • Sep 01 2016 14:40

    galthaus on fix-intermittent-network-role-creation-failure

    (compare)

rackneng
@rackneng
[wdennis] Sounds like the way to go...
[wdennis] Any docu on that process?
[zehicle] @wdennis we are creating a video for this. Our first attempt was pretty close, but needs to be updated for sound quality. https://youtu.be/yy7-2D4jXXg
[wdennis] Will check out... thx @zehicle
rackneng
@rackneng
[wdennis] OK, booted a new machine, only changed the name & profile, NO custom preseed/partitioning set, still getting the "No root file system is defined" error...
[wdennis] machine JSON
[wdennis] generated preseed
[wdennis] Can anyone here see any problems in the generated preseed? Should be DRP standard...
rackneng
@rackneng
[greg] You could try to boot into sledgehammer and check to see if the disk is already partitioned, if so, wipe it (like in the erase-disk task), and try. See if the LVM pre-existing is getting in the way.
[wdennis] I actually did do that (the disk was used, did have pre-existing LVM; I did a vgremove then pvremove and thereafter dd if=/dev/zero of=/dev/sda bs=512 count=1 to wipe MBR + part tbl
[wdennis] So to DRP should look like a blank disk.
rackneng
@rackneng
[greg] it isn’t DRP - it is ubuntu.
[wdennis] You are right
[wdennis] But anyways.
[greg] You may need to add the erase-hard-disks-for-os-install to your flow.
[greg] it does this:
#!/bin/bash
# Nuke it all.
declare vg pv maj min blocks name
# Make sure that the kernel knows about all the partitions
for bd in /sys/block/sd*; do
    [[ -b /dev/${bd##*/} ]] || continue
    partprobe "/dev/${bd##*/}" || :
done
# Zap any volume groups that may be lying around.
vgscan --ignorelockingfailure -P
while read vg; do
    vgremove -f "$vg" || :
done < <(vgs --noheadings -o vg_name)
# Wipe out any LVM metadata that the kernel may have detected.
pvscan --ignorelockingfailure
while read pv; do
    pvremove -f -y "$pv" || :
done < <(pvs --noheadings -o pv_name)
# Now zap any partitions along with any RAID metadata that may exist.
while read maj min blocks name; do
    [[ -b /dev/$name && -w /dev/$name && $name != name ]] || continue
    [[ $name = loop* ]] && continue
    [[ $name = dm* ]] && continue
    [[ $name = fd* ]] && continue
    mdadm --misc --zero-superblock --force /dev/$name || :
    if (( blocks >= 2048)); then
        dd "if=/dev/zero" "of=/dev/$name" "bs=512" "count=2048"
        dd "if=/dev/zero" "of=/dev/$name" "bs=512" "count=2048" "seek=$(($blocks - 2048))"
    else
        dd "if=/dev/zero" "of=/dev/$name" "bs=512" "count=$blocks"
    fi
done < <(tac /proc/partitions)
rackneng
@rackneng
[greg] We found you also have to blast the end of the disk.
[wdennis] Is that a new stage?
[greg] it is a task that can be added to stage.
[shane] LVM leaves nasty poo all over the place and is a nightmare to get rid of ...
[wdennis] Yup
[wdennis] But I thought vgremove/pvremove would get rid of it...
[shane] nope
[greg] nope.
[wdennis] And the "nope"s have it!
[wdennis] Why does it work tho when I re-install one of my DRP-installed hosts? B/c it has same LVM structure?
[greg] probably
[greg] I’ll probably need to post a new flow.
[greg] at some point that uses stage-chooser and a pre-stage to wipe the disk, but that is later.
rackneng
@rackneng
[wdennis] Booted the non-installing node with sledgehammer, and doing a dd if=/dev/zero of=/dev/sda bs=1M to wipe the disk..
[wdennis] Of course, it's a 2TB disk, so that'll run for a while...
[wdennis] We'll see what I get with the normal install thereafter
rackneng
@rackneng
[andreas.holmsten] @wdennis the no root file system is exactly what i had to troubleshoot as well. First off d-i partman/auto expert_recipe string in the partitioning scheme is incorrect syntax. It should be d-i partman-auto/expert_recipe string. Secondly I moved:
`\
d-i partman/confirm_write_new_label boolean true
d-i partman/choose_partition select finish
d-i partman/confirm boolean true
d-i partman/confirm_nooverwrite boolean true
[andreas.holmsten] The single reason the default preseed and partitioning work is because of the incorrect syntax. Since no manual scheme is selected the default atomic scheme is
rackneng
@rackneng
[andreas.holmsten] I've now created a PR: digitalrebar/provision-content#52
[andreas.holmsten] Observe that I'm not really that good with preseed so my assumptions might be wrong but it's what resolved the issue for me
rackneng
@rackneng
[greg] @andreas.holmsten seems expert to me :D
[greg] I’ll review and pull it in
Simon
@analbeard
hi all - currently kicking the tyres to see if Rebar fits my use-case - looking good so far! I did run into this whilst attempting to pxe my first victim: "[0:1]TFTP: lpxelinux.0: transfer error: sending block 0" and the issue error seems to come from here https://github.com/digitalrebar/provision/blob/master/midlayer/tftp.go#L68. unfortunately i'm not familiar with Go and i'm unsure as to why that error has cropped up - can it just not find lpxelinux.0?
rackneng
@rackneng
[wdennis] @andreas.holmsten Thanks - together maybe we can get a good base partitioning template, and then (my hope) maybe collaborate on community partitioning templates
[wdennis] Also @greg / @vlowther - maybe a good idea to put all of the partitioning (d-i partman*) directives into the base partitioning template - right now they are spread out over the preseed template and the partitioning template
rackneng
@rackneng
[shane] @analbeard we're in a meeting - but check that your DRP Endpoint doesn't have an asymmetric routing issue. You might try to add --static flag to DRP start up, with the IP address from the interface on the provisioning machine side
Simon
@analbeard
thanks Shane, that's certainly a possibility - the environment might need a little more work first
rackneng
@rackneng
[wdennis] Hey RackN folk - at an Ansible training today, they are still referencing Cobbler on their slide deck:
rackneng
@rackneng
[wdennis] Interesting - they keep mentioning Cobbler in their deck - strange for a dying OSS project to get such RHAT mentions…
[spector] The project is essentially “dead”. They release about 1 or 2 a year but it isn’t active at all
[wdennis] May be an oppt’y to reach out to RHAT folks and pitch your product (I’m sure you’ve probably thought of that already :) )
[spector] Yup, next time you see this go ahead and raise the hand and tell them all about Digital Rebar and to get with it!!!! We will of course reach out