Skip to content
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

runtime-cleanup: don't strip device-mapper-event #1328

Closed
wants to merge 1 commit into from

Conversation

AdamWill
Copy link
Contributor

The openQA install_blivet_lvmthin test - which, as the name implies, does an LVM thinp install - is consistently failing (only on ppc64le, for some reason). The logs show that lvm2 is trying to run dmeventd as part of some kind of check of the pool, but it's failing because dmeventd isn't there:

Failed to call the 'LvCreate' method on the '/com/redhat/lvmdbus1/ThinPool/0' object: GDBus.Error:org.freedesktop.DBus.Python.dbus.exceptions.DBusException: ('com.redhat.lvmdbus1.Lv', 'Exit code 5, stderr = /usr/sbin/dmeventd: stat failed: No such file or directory, /usr/sbin/dmeventd: stat failed: No such file or directory, Check of pool fedora/00 failed (status:64). Manual repair required!, Failed to activate thin pool fedora/00.')

So, apparently we do need this now. Whether to use dmeventd or not seems to be a build time check/config option for lvm2, but I don't think we can turn that off just to save a bit of space in the installer image - the package is only about 50K.

The openQA install_blivet_lvmthin test - which, as the name
implies, does an LVM thinp install - is consistently failing
(only on ppc64le, for some reason). The logs show that lvm2 is
trying to run dmeventd as part of some kind of check of the
pool, but it's failing because dmeventd isn't there:

Failed to call the 'LvCreate' method on the '/com/redhat/lvmdbus1/ThinPool/0' object: GDBus.Error:org.freedesktop.DBus.Python.dbus.exceptions.DBusException: ('com.redhat.lvmdbus1.Lv', 'Exit code 5, stderr = /usr/sbin/dmeventd: stat failed: No such file or directory, /usr/sbin/dmeventd: stat failed: No such file or directory, Check of pool fedora/00 failed (status:64). Manual repair required!, Failed to activate thin pool fedora/00.')

So, apparently we do need this now. Whether to use dmeventd or
not seems to be a build time check/config option for lvm2, but
I don't think we can turn that off just to save a bit of space
in the installer image - the package is only about 50K.

Signed-off-by: Adam Williamson <[email protected]>
@coveralls
Copy link

Pull Request Test Coverage Report for Build 5297278368

  • 0 of 0 changed or added relevant lines in 0 files are covered.
  • No unchanged relevant lines lost coverage.
  • Overall coverage remained the same at 47.495%

Totals Coverage Status
Change from base Build 5137526410: 0.0%
Covered Lines: 1701
Relevant Lines: 3429

💛 - Coveralls

@bcl
Copy link
Contributor

bcl commented Jun 20, 2023

It looks like this starts the service when it is installed so I'm a bit cautious about adding it. I'll try to do some tests on a new iso soon, but it might also be a good idea to ask @poncovka or @M4rtinK what Anaconda thinks about this.

@bcl
Copy link
Contributor

bcl commented Jun 21, 2023

Adding this turns on the dm-event.socket listener, and during the install I see logging about dmeventd trying to dlopen libdevmapper-event-lvm2thin.so which fails because we are removing lvm2-libs. It seems to fail gracefully, on x86_64 at least, and the install using lvm thinp finishes and boots.

But if we also add dm-event.socket dm-event.service to the systemctl disable in runtime-postinstall.tmpl it seems to be happy on x86 at least. The lvm2-libs package adds about 2M so if we can avoid adding that back in it would be nice.

bcl added a commit to bcl/lorax that referenced this pull request Sep 6, 2023
We don't install dmeventd or the event libraries so monitoring will fail
to start. The should solve weldr#1328 and rhbz#2236524

With this in place there are no syslog messages about dmeventd or event
libraries. lvm.conf does contain logged messages about dmeventd falling
back to various things, but no actual errors.
@bcl
Copy link
Contributor

bcl commented Sep 6, 2023

@AdamWill is there any way you could test #1343 with ppc64le?

@AdamWill
Copy link
Contributor Author

AdamWill commented Sep 6, 2023

Yes, but it requires a bit of bodging and it's not my number 1 priority right now, so I can't promise I'm gonna get to it. I will try, though!

@AdamWill
Copy link
Contributor Author

AdamWill commented Sep 7, 2023

BTW, the bug is actually fixed in Rawhide already since 2023-09-01, because Marian's fix for device-mapper-persistent-data landed that day ( device-mapper-persistent-data-1.0.6-2.fc40 ).

@AdamWill
Copy link
Contributor Author

AdamWill commented Sep 7, 2023

So, since the install fail is already fixed a different way on Rawhide, I can't immediately confirm that this would also have fixed it...but it at least doesn't break install again: https://openqa.stg.fedoraproject.org/tests/3141546

I can re-run the process on F39 tomorrow, I guess. F39 doesn't have the other fix yet, so we can see if this fixes the install there.

@bcl
Copy link
Contributor

bcl commented Sep 7, 2023

I can re-run the process on F39 tomorrow, I guess. F39 doesn't have the other fix yet, so we can see if this fixes the install there.

Thanks!

@bcl
Copy link
Contributor

bcl commented Sep 7, 2023

Merged the monitoring=0 change in #1343

@bcl bcl closed this Sep 7, 2023
@AdamWill
Copy link
Contributor Author

AdamWill commented Sep 8, 2023

Unfortunately seems like the bug still happens: https://openqa.stg.fedoraproject.org/tests/3146620 . I'll verify that the test was good tomorrow, but I think it was.

@bcl
Copy link
Contributor

bcl commented Sep 8, 2023

Hrm. So the error about dmeventd isn't there anymore. I see:

ERROR:anaconda.modules.storage.installation:Failed to create storage layout: Failed to call the 'LvCreate' method on the '/com/redhat/lvmdbus1/ThinPool/0' object: GDBus.Error:org.freedesktop.DBus.Python.dbus.exceptions.DBusException: ('com.redhat.lvmdbus1.Lv', 'Exit code 5, stderr = Check of pool fedora_fedora/00 failed (status:64). Manual repair required!, Failed to activate thin pool fedora_fedora/00.')

In the storage log for the failed test. But I can't tell what caused it.

@bcl bcl reopened this Sep 8, 2023
@AdamWill
Copy link
Contributor Author

AdamWill commented Sep 8, 2023

I guess that could still simply be caused by the bug the device-mapper-persistent-data update fixes? edit: I checked and the test does look valid - the lorax scratch build did apply the patch, and the openQA installer build test did use the scratch lorax.

@bcl
Copy link
Contributor

bcl commented Sep 28, 2023

Closing because it looks like this was fixed by https://bodhi.fedoraproject.org/updates/FEDORA-2023-b45924046a

@bcl bcl closed this Sep 28, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants