DATAPUISTOKEMISTI

I had unfortunate opportunity to play with RedHat / Fedora / CentOS distribution today. I have to say that as much as I hate Ubuntu, CentOS is even worse than it was few years ago when CentOS5 was still current. And now we're ignoring RHEL7 / CentOS7 where headless server installer requires GUI that's optimized for touch screen and tries to imitate Ipad. Recommended solution? Use VNC to connect installer. Aargh! I should probably have Ipad to run that VNC client - for improved user experience you know.

For 32-bit source to 64-bit see my older post . Old install was using software RAID-1, but did not have LVM. Small 200MB /boot partition, 8GB swap and rest as one root partition. We're redoing it completely so what it used to be doesn't really matter. All existing data on root will be lost.

# dnsmasq --version Dnsmasq version 2.68 Copyright (c) 2000-2013 Simon Kelley Compile time options: IPv6 GNU-getopt DBus i18n IDN DHCP DHCPv6 no-Lua TFTP conntrack ipset auth # md5sum dnsmasq* 8a704b6aa977df9485b3faf940cc5e35 dnsmasq_2.68-1_all.deb 60bafb9b863671bb02595505a447270c dnsmasq-base_2.68-1_amd64.deb There was odd " [icmp6 sum ok] ICMP6, destination unreachable, unreachable port " going out from my server every time DNS reply was received. Turns out it was because dnsmasq IPv6 support is buggy .

Headless Ubuntu 14.01 LTS server with full disk encryption, remote unlock over SSH, software RAID, LVM and support for over 2TB disks with EFI and BIOS MBR boot. That's what this post is about. Oh, and pics are at the end.

Dirty hack to renew DHCP lease on Cisco 881 if Internet access is lost. This can happen when Cisco has valid DHCP lease from ISP and then cable modem or DSL router is power cycled but switch between Cisco and upstream device keeps link-state up. Options are either to fix it manually (pull cable / change config / reboot Cisco) or hack something like below. I'm also pinging over VPN tunnel to Intranet as it would be shame to break this due Level 3 and Google blocking ICMP which might happen one day.

Windows GUI prevents you from setting this with helpful error message of "All of the bits in the host address portion of the IP address are set to 0". IP stack seems to be fine with /31 networks so just set it from CLI. netsh interface show interface netsh interface ip set address name="Local Area Connection" source=static addr=10.71.219.238 mask=255.255.255.254 gateway=10.71l.219.239

Trying to move virtual machine between datastores (local or remote) can be painful without vCenter and SVmotion. Easy and fast way is to use traditional VI Client to move contents of virtual machine directory via datastore browser (Move To feature). Processing stays within ESXi and goes quickly. VMware converter could also be used, but it's SLOOOOOW and everything goes via management PC which is dumb design. GhettoVCB restores aren't fastest either. This is quite obvious workaround, but I have to admit didn't realize this until today. Since backups are NFS mounted I can simply import VMX file from backup directory using datastore browser. Yes, this will result in changes to backup which is not that great, but at least it's quick - even biggest VM can be restored in couple minutes. And that issue with changes to backup could be easily solved with ZFS snapshots.