statbugs1001 or just staticbugs or something

there will very likely just be a single host so staticbugs.eqiad.wmmet or something would be best following the misc naming style as this isn't really a 'service' perse.

that said, "misc naming style" would mean to use element names in eqiad and star names in codfw :)

..and then have service names that are not equal to the hostname.. which is ok

I am fine with either approach. Wanna propose a name ?

I propose staticbugs.eqiad.wmnet simply because the 100x part makes it look ugly and it won't have another host unless this somehow becomes more popular than actual wikis :)

The 100x part indicates site as well as member in a series.

staticbugs.eqiad.wmnet is against the policy. TL;DR It's either element/star (eqiad/codfw respectively) or serviceXNNN where X=[1,2,3,4] for eqiad, codfw, esams,ulsfo.

Next element in line then I guess is my view point though I'm not against the staticbugs1001 name if that's what seems best?

That sounds like a bit of a silly policy to me...

For the interested, this is documented in:

https://wikitech.wikimedia.org/wiki/Infrastructure_naming_conventions

@JohnLewis: Next element in line it is, then

If this is a VM, and not bare metal, it shouldnt use up element names. The element names are allocated for bare metal misc servers.

This was an ongoing discussion on 'how do we name misc vms' that I didn't think had been resolved. I'm against them sharing the naming scheme of bare metal misc servers.

If its a VM, I think it should be named by its role and all VMs should have some designation in their hostnames to denote they aren't on bare metal.

I'd call this vm something like: staticbugzilla1001, rather than staticbugs, since bugs can be ANY bug tracker.

Infrastructure naming convention discussions are the worst.

There is no really nice way to name these in regards to misc bare-metal versus misc vm, considering their use cases will overlap. As such, Alex has convinced me to change stance and revoke my statement not to use element names/star names for misc vm servers.

Change 222149 had a related patch set uploaded (by Alexandros Kosiaris):
Introducing bromine.eqiad.wmnet

https://gerrit.wikimedia.org/r/222149

Change 222149 merged by Alexandros Kosiaris:
Introducing bromine.eqiad.wmnet

https://gerrit.wikimedia.org/r/222149

bromine.eqiad.wmnet is ready on ganeti01.svc.eqiad.wmnet with MAC: aa:00:00:f1:36:3a. DHCP configuration and installation are not done yet.

Change 222172 had a related patch set uploaded (by Dzahn):
add bromine to dhcp

https://gerrit.wikimedia.org/r/222172

Change 222172 merged by Dzahn:
add bromine to dhcp

https://gerrit.wikimedia.org/r/222172

merged DHCP config (thanks John)

tried to get console:

[ganeti1003:~] $ sudo gnt-instance console bromine.eqiad.wmnet
2015/07/01 22:03:13 socat[47275] E connect(5, AF=1 "/var/run/ganeti/kvm-hypervisor/ctrl/bromine.eqiad.wmnet.serial", 64): No such file or directory

                                                                   Connection to ganeti1003.eqiad.wmnet closed.

Failure: command execution error:
Connection to console of instance bromine.eqiad.wmnet failed, please check cluster configuration

after waiting a bit and restarting it i got console and saw the installer.
then, as with planet1001 at first:

    ┌────────┤ [!!] Install the GRUB boot loader on a hard disk ├─────────┐
    │                                                                     │
    │                      GRUB installation failed                       │
    │ The 'grub-pc' package failed to install into /target/. Without the  │
<Tab> moves; <Space> selects; <Enter> activates buttons───────────────────┘

after that the installation finished and the system powered down.

when i powered it up again the installer started from scratch

next attempt the installer finished without these error messages above. after finishing it shuts down the machine.

after you power it up again, installing starts again.

sure i am not supposed to manually switch the boot order (as i tried with planet1001)

In T103604#1418961, @Dzahn wrote:

merged DHCP config (thanks John)

tried to get console:

[ganeti1003:~] $ sudo gnt-instance console bromine.eqiad.wmnet
2015/07/01 22:03:13 socat[47275] E connect(5, AF=1 "/var/run/ganeti/kvm-hypervisor/ctrl/bromine.eqiad.wmnet.serial", 64): No such file or directory

                                                                   Connection to ganeti1003.eqiad.wmnet closed.

Failure: command execution error:
Connection to console of instance bromine.eqiad.wmnet failed, please check cluster configuration

That means the VM was powered off

In T103604#1419079, @Dzahn wrote:

after waiting a bit and restarting it i got console and saw the installer.
then, as with planet1001 at first:

    ┌────────┤ [!!] Install the GRUB boot loader on a hard disk ├─────────┐
    │                                                                     │
    │                      GRUB installation failed                       │
    │ The 'grub-pc' package failed to install into /target/. Without the  │
<Tab> moves; <Space> selects; <Enter> activates buttons───────────────────┘

This is the second time this issue is encountered. I 'd like to debug it but alas, it resolved itself on the next reboot

In T103604#1419478, @Dzahn wrote:

next attempt the installer finished without these error messages above. after finishing it shuts down the machine.

after you power it up again, installing starts again.

sure i am not supposed to manually switch the boot order (as i tried with planet1001)

You are supposed to do exactly that. https://wikitech.wikimedia.org/wiki/Ganeti#Set_boot_order_to_disk

This is a limitation of the current version of the Ganeti software and is expected to be solved (upstream is aware)

I did that and set the boot order to disk during install. After shutdown i started it up again.

Then i tried to connect to console.. no output

all that said, it still worked and i could ssh to it with the "new_install" key and it's up and running :)

a couple minutes later the screen with the "frozen" console came to life and works now:

bromine login:
Debian GNU/Linux 8 bromine ttyS0