Question about GHOST vulnerability
Among the things they say are:
If the version of eglibc matches, or is more recent than, the ones listed here, you are safe from the GHOST vulnerability:
Ubuntu 12.04 LTS: 2.15-0ubuntu10.10
Ubuntu 10.04 LTS: 2.11.1-0ubuntu7.20
Debian 7 LTS: 2.13-38+deb7u7
In our case we are using Ubuntu 10.10 and our version of eglibc is "Ubuntu EGLIBC 2.12.1-0ubuntu10.4) 2.12.1".
Is our server vulnerable or safe? Yes, I know we really should upgrade to a newer distribution and are planning to upgrade to 14.04 LTS, but that is quite time-consuming for all our servers. So in the meantime, I was wondering if we were vulnerable or not. And if so, could we just upgrade glibc and reboot in the meantime?
Thanks,
doug
19 Replies
not vulnerable
So if that is true we are ok with regards to this particular vulnerability for now.
doug
Given how long Ubuntu 10.10 has been unsupported, you're likely vulnerable to many other things, like Shellshock and a dozen OpenSSL issues.
Yes, you can "just" upgrade glibc. Just download the eglibc source package, backport and apply the patch for the vulnerability, and build a new package. The time spent learning how to do that might be better used upgrading to a supported distro.
Edit: synapt: That's not how it works. Ubuntu backported the patch to 10.04's eglibc 2.11.1 package, because it's still supported. They haven't backported it to 10.10 since it's long unmaintained.
But what about the fact that our running of the GHOST test as described at
Is that a valid test?
Thanks,
doug
@mnordhoff:
Edit: synapt: That's not how it works. Ubuntu backported the patch to 10.04's eglibc 2.11.1 package, because it's still supported. They haven't backported it to 10.10 since it's long unmaintained.
Doh, completely mis-read 10.10, ignore me
So at the minimum we have to upgrade glibc.
doug
I would, of course, like to upgrade all our Ubuntu 10.10 servers to Ubuntu 14.04 LTS. And I hope to do that as soon as possible. But it seems like such a mammoth undertaking and I have no experience upgrading a distribution like that.
I know the general idea: create a new Linode with the Ubuntu 14.04 LTS distribution; use rsync (or possibly just scp) to copy over a few critical areas, such as our data and hopefully the users and their directories; shut down both servers and swap the IPs; then bring up the new linode and make sure that everything is running on the ok before shutting down the old linode.
But it seems pretty time consuming (unless there is a simpler process than I described) because we have a lot of linodes. That's why I was thinking maybe just a glibc upgrade in the meantime might be a good first step if this is a very dangerous vulnerability. Just to get it patched in the meantime before we do the distro upgrades. Does that make sense?
But our distro is Ubuntu 10.10 and is no longer supported, so I don't even know for sure what the best glibc is for that for this purpose.
Thanks,
doug
Thanks,
doug
Actually, with apt-get dist-upgrade I got some upgrade messages like this:
WARNING: The following packages cannot be authenticated!
nagios-nrpe-server tzdata ntpdate initramfs-tools initramfs-tools-bin sysvinit-utils sysv-rc initscripts
openssh-server openssh-client python-apt gdb nagios-nrpe-plugin ntp
Install these packages without verification [y/N]? N
I wasn't sure about that "without verification" message so I entered N for now.
doug
Support has always been very friendly to me. But I'm always polite.
They suggested not upgrading in place though because apparently, while it works most of the time, they have seen cases where the linode won't boot afterwards. What do you think?
Obviously if we can upgrade in place to any supported version that would be a huge huge huge time saver.
Thanks,
doug
@obs:
You could clone the node to a new node then upgrade the clone in place, if it works great if it doesn't you still have the old node to run off while you build a new one from scratch.
Yes, that's what I was thinking I would try.
The "uncertainty factor" here is that I have about 40 linodes. So I'm a bit worried that I do it with one as a test and it works, but then along the way "linode 27" doesn't work. I suppose if I take a snapshot backup before each upgrade and it doesn't work I can always restore from the snapshot.
Do you think the
do-release-upgrade
mentioned at
is the way to try this?
Thanks,
doug
Or if you really want to do something sane then you should script your setup procedure using something like ansible/chef/puppet (or even bash scripts will do). That way you have a verified way of quickly setting up nodes. To get to 12.04 you have to go via 11.04, 11.10 then to 12.04 you can only jump from LTS release to another LTS release i.e. 10.04 > 12.04 > 14.04.
You should stick to 12.04 or 14.04 since they're supported for 5 years from the date of release.
Whatever you do it's going to be a chore! Good luck!
What about the in-place upgrade from 10.10 to 12.04 mentioned at the askubuntu.com site listed above? No good?
Thanks,
doug
@douglerner:
What about the in-place upgrade from 10.10 to 12.04 mentioned at the askubuntu.com site listed above? No good
It's not supported or recomended, the official way is to upgrade through each release.
TBH if I was in your shoes I'd replace all the servers with fresh Ubuntu 14.04 servers, the OS is supported until 2019 and then you'd have a nice clean server and you know what's on it without having to worry about any weirdness from anything you inherited.