Difference between revisions of "HP DL360 G7 Hypervisor"

From W9CR
Jump to navigation Jump to search
Line 356: Line 356:
 
== Fix the /etc/pve/storage.cfg ==
 
== Fix the /etc/pve/storage.cfg ==
  
−
vim /etc/pve/storage.cfg  
+
vim /etc/pve/storage.cfg  
−
Move MoleMan from local-zfs to localDataStore
+
Move MoleMan from local-zfs to localDataStore
 +
 
 +
= Reboot =
 +
Reboot and check everything is working on the cluster and move VM's

Revision as of 22:55, 3 September 2024


This describes the build process for Proxmox 8.2 on existing Compaq/HP DL360 G7 Server

These have served me well for several years and while the could stand to be upgraded, they are maxed out on CPU, RAM and disk, and have plenty of life left in them. As I'd installed Proxmox years ago, and before I knew what I know now regarding the naming local storage for ZFS pools, I decided to rebuild them during the 8.2 upgrade.

Config

The Servers consist of the following:

  • 2x Xenon X5675 @3.07 GHz (12 real/24 virtual cores)
  • 192 GB (12x16gb DDR3 1600 MHz) RAM
  • 8x SAS 1.6 TB SSD HUSMM141CLAR1600
  • LSI SAS2308 storage controller - Firmware link
  • Mellanox MT27520 ConnectX-3 Pro dual 10g Nic

Disk Layout

I use local ZFS storage on most of my HVs, and as such proxmox cannot migrate or replicate VMs or CTs from nodes with different pool names. By default everything is rpool, and rpool is the root disks, which on other newer servers are used for booting only. This means that one of the core features of proxmox will not function unless it's fixed.

As I'm not wanting to invest more in these servers and figure out how to put a boot disk in, we must boot off the existing SSD's. This has the limitation of requiring 512b sector disks and GRUB as the servers are too old to use UEFI boot and native 4k disks. Proxmox does write to it's root disk so vmware kinda boot (ie USB boot or DOM SATA) will not work, but the IO requirements are not that intense. The decision was made to use part of the disks as a raidz3 pool with 16gb per disk making 80gb reserved for the root device.

The localDataStore pool will be raidz1 with the remaining space on each device. zraid1 was chosen as each VM is fully backed up and even if >1 disks fail, the server will still boot as we can lose 3 disks before it's unbootable. I've not lost a disk on these servers before either, and cold spares are on site.

We'll also pad out the space between the root pool partition and the localDataStore partitions with a 16GiByte partition, allowing easy expansion of the root pool if needed in the future. Finally we'll leave 16MiB of space at the end of each disk.

Thus we'll have the following: 

Disk /dev/sda: 3125627568 sectors, 1.5 TiB
Model: HUSMM141CLAR1600
Sector size (logical/physical): 512/4096 bytes
Disk identifier (GUID): 7E89E66A-F81F-431A-B71A-F8BA7583CDC1
Partition table holds up to 128 entries
Main partition table begins at sector 2 and ends at sector 33
First usable sector is 34, last usable sector is 3125627534
Partitions will be aligned on 8-sector boundaries
Total free space is 32774 sectors (16.0 MiB)

Number  Start (sector)    End (sector)  Size       Code  Name
   1              34            2047   1007.0 KiB  EF02
   2            2048         2099199   1024.0 MiB  EF00
   3         2099200        33554432   15.0 GiB    BF01
   4        33554440        67108871   16.0 GiB    BF01  Solaris /usr & Mac ZFS
   5        67108872      3125594767   1.4 TiB     BF01  Solaris /usr & Mac ZFS

# zpool list -v
NAME                               SIZE  ALLOC   FREE  CKPOINT  EXPANDSZ   FRAG    CAP  DEDUP    HEALTH  ALTROOT
localDataStore                    11.4T  2.27T  9.12T        -         -     0%    19%  1.00x    ONLINE  -
  raidz1-0                        11.4T  2.27T  9.12T        -         -     0%  19.9%      -    ONLINE
    scsi-35000cca0504cf90c-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca05053df48-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504d201c-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0506d06e0-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504d3378-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0506a76bc-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504c33c8-part5  1.42T      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504d1e3c-part5  1.42T      -      -        -         -      -      -      -    ONLINE
rpool                              119G  66.4G  52.6G        -         -     1%    55%  1.00x    ONLINE  -
  raidz3-0                         119G  66.4G  52.6G        -         -     1%  55.8%      -    ONLINE
    scsi-35000cca0504cf90c-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca05053df48-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504d201c-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0506d06e0-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504d3378-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0506a76bc-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504c33c8-part3  15.0G      -      -        -         -      -      -      -    ONLINE
    scsi-35000cca0504d1e3c-part3  15.0G      -      -        -         -      -      -      -    ONLINE

Preparation

As these are currently running in the pool, we need to:

  • Migrate VM's off of the HV
  • Ensure shutdown VMs are backuped in PBS
  • Backup the remaining rpool
  • Check that the ILO works

ZFS backup to FatTony

Backup the pve to /root/pve 

cp -av /etc/pve/ /root/pve-etc

Make a pool to store this on fattony 

ssh root@192.168.8.184 zfs create testpool/`hostname`-old

Make a snapshot of the rootfs 

zfs snapshot rpool/ROOT/pve-1@20240903-01

Send it to the host 

zfs send rpool/ROOT/pve-1@20240903-01 |pv| ssh -c aes128-gcm@openssh.com 192.168.8.184 zfs recv -Fdu -o canmount=noauto testpool/`hostname`-old


Do the Install

Put the 8.2 iso on the server and connect it to the ILO CD drive http://192.168.8.150/~bryan/zfs/proxmox-ve_8.2-2.iso

This may take 5-10 min to load, look the the http server logs.

At the GRUB prompt you will need to add 

video=1024x768@60 nomodeset

after the linux line in the gui installer

'e' then edit as below and hit ctrl-x <image>

accept the GNU license

Setup the Disks

  • zfs RAIDZ-3
  • all disks
  • ashift = 12
  • compress on
  • checksum on
  • copies 1
  • ARC max size 16384
  • hdsize 16.0 GB


'set the county and time zone

  • United states
  • America/New_york (you need to use arrow keys, using the mouse doesn't work)
  • Keyboard U.S. English

set root passwd and email

setup the network 

we have to manually set this as it needs a vlan tag.

Install

  • reboot when done
  • don't forget to remove the CD

post reboot

network

The nics are on bonding so we need to setup LACP/bonding 

brctl delif vmbr0 ens2
ip link add bond0 type bond mode 802.3ad
ip link set ens2 down 
ip link set ens2 master bond0
ip link set ens2d1 master bond0
ip link add link bond0 name bond0.42 type vlan id 42
ip link set bond0 up
ip link set bond0.42 up
brctl addif vmbr0 bond0.42

The network should be up and you can use ssh from here

Now push the old config from FatTony 

scp -o UserKnownHostsFile=/dev/null /etc/hosts Moleman:/etc/hosts 
scp -o UserKnownHostsFile=/dev/null /testpool/MoleMan-old/ROOT/pve-1/etc/ssh/ssh_host_* MoleMan:/etc/ssh/

restart ssh on molman 

scp /root/.ssh/authorized_keys root@MoleMan://root/.ssh/authorized_keys  
scp /testpool/MoleMan-old/ROOT/pve-1/etc/network/interfaces root@MoleMan:/etc/network/interfaces  
scp /testpool/MoleMan-old/ROOT/pve-1/etc/udev/rules.d/70-persistent-net.rules root@MoleMan:/etc/udev/rules.d/
scp /testpool/MoleMan-old/ROOT/pve-1/etc/network/if-up.d/00-keekles.sh MoleMan:/etc/network/if-up.d/00-keekles.sh

Reboot now verify the network is up

Add Software

Configure the repos 

mv /etc/apt/sources.list.d/ceph.list /etc/apt/sources.list.d/ceph.list.bak
mv /etc/apt/sources.list.d/pve-enterprise.list /etc/apt/sources.list.d/pve-enterprise.bak
echo -e "deb http://download.proxmox.com/debian/pve bookworm pve-no-subscription \ndeb http://download.proxmox.com/debian bookworm pve-no-subscription" >/etc/apt/sources.list.d/pve-no-subscription.list 



apt-get update 
apt-get install sg3-utils-udev sdparm ledmon lsscsi net-tools nvme-cli lldpd rsyslog ipmitool vim unzip git fio sudo locate screen snmpd libsnmp-dev mstflint pv

Run through the standard configs

https://wiki.w9cr.net/index.php/Hypervisor_1RU_2024#Software

Add users

useradd -s /bin/bash -m -U -G sudo bryan  
useradd -s /bin/bash -m -U -G sudo stacy
echo "bryan ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers.d/00-admins  
echo "stacy ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers.d/00-admins

From Fattony 

rsync -avz /testpool/carbonrod-old/ROOT/pve-1/home/bryan/.ssh root@moleman:/home/bryan/  
rsync -avz /testpool/carbonrod-old/ROOT/pve-1/home/stacy/.ssh root@moleman:/home/stacy/


Configure the MTA

Configure Observium

This has mutiple parts, and the hardest is the config of the hp tools for the ILO3. This requires a dummy package and some other custom links to get stuff working.

As part of the SNMP config we'll be making this snmpv3 only.

  • Install needed packages
sudo apt-get install snmpd php perl curl xinetd snmp libsnmp-dev libwww-perl rrdtool mailgraph
dpkg-reconfigure mailgraph
  • install the current distro program on the target:
 sudo curl -o /usr/local/bin/distro https://gitlab.com/observium/distroscript/raw/master/distro
 sudo chmod +x /usr/local/bin/distro
  • Do this from eyes
export SERVER=<hostname>

scp /opt/observium/scripts/observium_agent_xinetd $SERVER:/etc/xinetd.d/observium_agent_xinetd
scp /opt/observium/scripts/observium_agent $SERVER:/usr/bin/observium_agent
ssh $SERVER mkdir -p /usr/lib/observium_agent
ssh $SERVER mkdir -p /usr/lib/observium_agent/scripts-available
ssh $SERVER mkdir -p /usr/lib/observium_agent/scripts-enabled
scp /opt/observium/scripts/agent-local/* $SERVER:/usr/lib/observium_agent/scripts-available
  • on target enable the various allowed items
ln -s /usr/lib/observium_agent/scripts-available/os /usr/lib/observium_agent/scripts-enabled
   #ln -s /usr/lib/observium_agent/scripts-available/zimbra /usr/lib/observium_agent/scripts-enabled
ln -s /usr/lib/observium_agent/scripts-available/dpkg /usr/lib/observium_agent/scripts-enabled
ln -s /usr/lib/observium_agent/scripts-available/ntpd /usr/lib/observium_agent/scripts-enabled
ln -s /usr/lib/observium_agent/scripts-available/virt-what /usr/lib/observium_agent/scripts-enabled 
ln -s /usr/lib/observium_agent/scripts-available/proxmox-qemu /usr/lib/observium_agent/scripts-enabled
ln -s /usr/lib/observium_agent/scripts-available/postfix_mailgraph /usr/lib/observium_agent/scripts-enabled
  • Edit /etc/xinetd.d/observium_agent_xinetd so the Observium server is allowed to connect. You can do this by substituting 127.0.0.1, or place your IP after it, separated by a space. Make sure to restart xinetd afterwards so the configuration file is read.
sudo service xinetd restart
  • Test from eyes
telnet $SERVER 36602

Setup SNMP and HP utils

  • configure the HP 10.80 repo for the HPutils
echo "deb http://downloads.linux.hpe.com/SDR/repo/mcp bionic/10.80 non-free" > /etc/apt/sources.list.d/hp-mcp.list
curl -sS https://downloads.linux.hpe.com/SDR/hpPublicKey2048.pub | gpg --dearmor > /etc/apt/trusted.gpg.d/hpPublicKey2048.gpg
curl -sS https://downloads.linux.hpe.com/SDR/hpPublicKey2048_key1.pub | gpg --dearmor > /etc/apt/trusted.gpg.d/hpPublicKey2048_key1.gpg
curl -sS https://downloads.linux.hpe.com/SDR/hpePublicKey2048_key1.pub | gpg --dearmor > /etc/apt/trusted.gpg.d/hpePublicKey2048_key1.gpg
  • install the libc6-i386 equivs package from carbonrod
scp /root/libc6-i686_2.36_all.deb moleman:/root/

Now on the host 

apt-get install libc6-i386
dpkg -i ./libc6-i686_2.36_all.deb
  • Install the HP utils
apt-get update
apt-get install hp-snmp-agents hp-health
  • The following symlinks are needed for the older version to work.
ln -s /usr/lib/x86_64-linux-gnu/libnetsnmpmibs.so.40 /usr/lib/x86_64-linux-gnu/libnetsnmpmibs.so.30  
ln -s /usr/lib/x86_64-linux-gnu/libnetsnmpagent.so.40 /usr/lib/x86_64-linux-gnu/libnetsnmpagent.so.30  
ln -s /usr/lib/x86_64-linux-gnu/libnetsnmp.so.40 /usr/lib/x86_64-linux-gnu/libnetsnmp.so.30
  • copy over the old snmpd.conf from FatTony
scp /testpool/MoleMan-old/ROOT/pve-1/etc/snmp/snmpd.conf  root@moleman:/etc/snmp/snmpd.conf
  • edit the snmpd.conf on MoleMan

Remove the 'rocommunity line' 

vim /etc/snmp/snmpd.conf
  • Make a SNMP User
sudo service snmpd stop
sudo service hp-health stop
sudo service hp-asrd stop
sudo service hp-snmp-agents stop
sudo net-snmp-config --create-snmpv3-user -ro -A keeklesSNMPpasswd -a SHA -x AES -X keeklesSNMPpasswd KeeklesSNMP
sudo service hp-health start
sudo service hp-asrd start
sudo service hp-snmp-agents start
sudo service snmpd start
  • test it from eyes
snmpwalk -v3 -u KeeklesSNMP -l authPriv -a SHA -A keeklesSNMPpasswd -x aes -X keeklesSNMPpasswd MoleMan 1.3.6.1.4.1.232

localDataStore config

Configure the disks

For sda - sdh we need to add some partitions. We must have them alligned on the 4096 byte boundary. 

Total free space is 32774 sectors (16.0 MiB) 

Number  Start (sector)    End (sector)  Size       Code  Name
   1              34            2047   1007.0 KiB  EF02
   2            2048         2099199   1024.0 MiB  EF00
   3         2099200        33554432   15.0 GiB    BF01
   4        33554440        67108871   16.0 GiB    BF01  Solaris /usr & Mac ZFS
   5        67108872      3125594767   1.4 TiB     BF01  Solaris /usr & Mac ZFS

1024*1024*16/512 = 32768 sectors from end 3125627534 - last sector end should be 67108872 - 3125594767 =3,058,485,895 sectors =

  • install partprobe
apt-get install parted
partprobe

Configure the ZFS for localDataStore

zpool create -f -o ashift=12 -O compression=lz4 -O atime=off -O xattr=sa localDataStore \
raidz1 /dev/sda5 /dev/sdb5 /dev/sdc5 /dev/sdd5 /dev/sde5 /dev/sdf5 /dev/sdg5 /dev/sdh5

zpool export localDataStore
zpool import -d /dev/disk/by-id/ localDataStore

zfs create localDataStore/proxmox

zfs snapshot rpool/var-lib-vz@xfer  
zfs send -vR rpool/var-lib-vz@xfer | zfs receive -Fdu localDataStore  
zfs umount rpool/var-lib-vz  
zfs destroy -r rpool/var-lib-vz  
zfs mount localDataStore/var-lib-vz  
zfs destroy -v localDataStore/var-lib-vz@xfer  
zfs set mountpoint=none rpool/data  
zfs set mountpoint=/localDataStore/proxmox localDataStore/proxmox
  • reboot the node now

Add node to cluster

  • from a node in the cluster delete the old node
pvecm delnode MoleMan
  • now add the node
pvecm add fink

Please enter superuser (root) password for 'fink': **********
Establishing API connection with host 'fink'
The authenticity of host 'fink' can't be established.
X509 SHA256 key fingerprint is D2:5C:21:E4:CE:8B:3A:6C:27:31:79:B3:52:5A:FE:83:3B:8F:8C:A2:28:F2:4D:E1:E6:DB:03:14:40:BA:F1:B6.
Are you sure you want to continue connecting (yes/no)? yes
Login succeeded.
check cluster join API version
No cluster network links passed explicitly, fallback to local node IP '192.168.8.181'
Request addition of this node
Join request OK, finishing setup locally
stopping pve-cluster service
backup old database to '/var/lib/pve-cluster/backup/config-1725421258.sql.gz'
waiting for quorum...OK
(re)generate node files
generate new node certificate
merge authorized SSH keys
generated new node certificate, restart pveproxy and pvedaemon services
successfully added node 'MoleMan' to cluster.

Fix the /etc/pve/storage.cfg

vim /etc/pve/storage.cfg 
Move MoleMan from local-zfs to localDataStore

Reboot

Reboot and check everything is working on the cluster and move VM's

Retrieved from "https://wiki.w9cr.net/index.php?title=HP_DL360_G7_Hypervisor&oldid=8078"