[Savannah-hackers-public] Savannah VM Maintenance Happening

[Bob Proulx]

Hello Savannah Hackers,

Ineiev, This might affect you if things get snarled up!  I'll try to
push these through with the least amount of downtime.

First let me apologize for being distracted with other things and
having little time recently for Savannah issues.  I just returned from
a week of being completely disconnected on a bicycle adventure with a
bicycle tour across Wyoming for example.  This summer has been very
busy!  I have not been able to keep up with the mailing lists.  But I
have carved out time to work on system upgrades this week and am
working through them.

Today I have been pushing security upgrades through.  These don't
radically change anything.  So nothing to note.  However I plan to
reboot the servers for the new kernels.  And that's possibly going to
be a problem.  And might possibly affect things today/tomorrow as we
(FSF admin and I) work through the problems.

Reboots are the topic of this message.  Last month we rebooted one of
the systems (internal2) and it did not boot.  This exposed a problem
with the way the VMs are hosted, booted, kernels selected, and so
forth.  This next is an "as I understand it" description second hand.
Ian has the authoritative view.

Currently the community0p server is configured to boot a private grub
and that private grub is configured to boot a specific boot path on
the VM.  This is due to the use of encrypted file system storage which
needs to be decrypted before use.  Which means that the two parts must
work together.  In the last month or so a change had been made in the
private grub configuration side to use /boot/vmlinuz instead of
/vmlinuz and that path not existing the system failed to boot.  This
was sorted out (Thanks Ruben!) and the system brought back online.
This event alerted us to this unexpected snarl.

So here we are now working through the details.  We are updating the
VMs hosted on the community server.  The current goal is to have all
of the grub configurations there boot the /boot/vmlinuz path and have
that path automatically updated (/etc/kernel-img.conf) when new
kernels are installed.

That's the plan!  We will be successful when the VMs once again reboot
cleanly.  It will probably take some tinkering.  I am starting with
the non-production VMs (internal2, frontend2) first as those don't
affect users.  But of course the goal is to have things working
smoothly for the production servers.

Bob

signature.asc
Description: PGP signature