New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
recover from failure (update to fedora 31) #1684
Comments
I opened a fedora bug for this: https://bugzilla.redhat.com/show_bug.cgi?id=1767743 |
Please give precise behavior when attempting to start w/ systemd, thanks! EDIT: Probably unnecessary, I see there is much more information in bz. |
I've performed a basic test of creating a stratis volume on fedora 30 then did an upgrade to fedora 31. There was a minor hang on first boot of the system (not sure what the cause of it was), but rebooted again and the volume mounted fine with all of its data. This test didn't involve LUKS. I'll review the sosreport in bz 1767743. |
From comment 3 of https://bugzilla.redhat.com/show_bug.cgi?id=1767743 there is an "os error 13" reported I believe there was a fix to the lock file location for stratis recently because of a different boot issue. Could this be a similar issue? Still looking... |
aanno, is stratisd running after boot, if not, can you try to start it? # systemctl enable stratisd;systemctl start stratisd ( please provide output) |
Dear drckeefe, thank you for looking into this. On the effected system, stratisd is not running after boot and I can't start it because of https://bugzilla.redhat.com/show_bug.cgi?id=1767773 . However, if I first I don't imply that every update from fedora 30 to fedora 31 is effected. My update was a bit rough because I had to disable some third party repos after the update to regain a usable system. Hence it is possible that all kind of evil has effected my update. However, as far I can tell my system is now in a clean (i.e. fedora 31 repos only plus nvidia driver) state. This ticket is more on my options for recovery. For me it feels a bit problematic that interaction with stratisd is based on DBus - as DBus might not be available in case of a disaster. For example, it is not available in the (fedora) emergency shell. In summary, stratisd seems not to be the reason for my ongoing problem with the system. There is no data loss, however at present the system is hardly usable. The problem at hand is caused by some strange interaction between selinux and stratisd. So probably only fedora users could encounter the problem... |
Versions affected: # rpm -qa selinux-policy* stratis*
stratis-cli-1.0.4-1.module_f31+6320+bf3c8975.x86_64
selinux-policy-targeted-3.14.4-39.fc31.noarch
selinux-policy-3.14.4-39.fc31.noarch
stratisd-1.0.5-1.module_f31+6320+bf3c8975.x86_64 |
Thank you for digging into this further and provide feedback to the issue it is very helpful information. I'm interested to know what your thoughts are on a quick process to reproduce this, such as:
Do you believe that the upgrade process impacts this configuration? Could this happen if a Stratis filesystem was created directly from Fedora 31? |
Thanks for the information aanno. Here was my process to get stratis to fail after installing stratis-cli
Installing: Now check the status of startisd
Nov 04 15:11:16 localhost-live systemd[1]: Started A daemon that manages a pool of block devices to create flexible file systems. Try to create a pool results in a failure because stratisd is not running Most likely stratis is unable to connect to the stratisd D-Bus service. Set selinux to 0 resolves the issue [root@localhost-live liveuser]# setenforce 0 Nov 04 15:13:58 localhost-live systemd[1]: Started A daemon that manages a pool of block devices to create flexible file systems. [root@localhost-live liveuser]# stratis pool create p1 /dev/vda |
Here are the journal messages related to stratisd being blocked by selinux Nov 04 15:33:25 localhost-live systemd[1]: Started A daemon that manages a pool of block devices to create flexible file systems. |
Very well, you have reproduced the problem! |
Having the same problem, clean fedora 30 to 31 upgrade for latest stratisd and will reproduce. Also, when I run dnf update, I get conflicting request : Trying out nightly :
And it works now ! Well, the service starts... EDIT :
Thanks |
Current status: The issue has been handled with https://bugzilla.redhat.com/show_bug.cgi?id=1755396 . However, the selinux problem is still there. Currently, there seems to be no attempt in the Fedora project to rectify the issue. |
The issue has finally been solved with https://bugzilla.redhat.com/show_bug.cgi?id=1794645 |
Yesterday, I updated my PC from fedora 30 to 31. The update turned out to be problematic, and now I could only boot into the 'emergency shell'.
Fedora installation disk layout is like this: All partitions are LUKS encrypted, root (
/
) is (directly) on LVM on a SSD./home
and/opt
are stratis managed (HDDs partition SSD cached).Investigations showed the following: boot into 'emergency shell' is because
/home
could not be mounted. However, I have problems to recover stratis both from the 'emergency shell' and from the fedora 31 live cd.In 'emergency shell' stratisd warns that there is no DBus. Hence the
stratis
cli is on no use. I have not found a way to successfully start DBus in the 'emergency shell'.From the live cd, I could install
stratisd
andstratis-cli
. But it is not possible to startstratisd
with systemd (some IO Error, permission denied).In chroot from live cd to my installed root (
/
) it is not possible to startstratisd
because systemd does not allow this in chroot.So, has somebody an idea what to do to recover the system?
(Cross posted on mailing list as well)
The text was updated successfully, but these errors were encountered: