Bug 1659656

The scriptlet is running:

usermod %{updateuser} -a -G virusgroup

The guidelines for users and groups (https://docs.fedoraproject.org/en-US/packaging-guidelines/UsersAndGroups/) does not directly address calling usermod, but the useradd examples do not redirect stderr - presumably because it is useful at times.

I'm guessing that usermod is calling sss_cache.  I'm not really sure what should be done here.  Let's see if the SSSD folks have an opinion about whether generating this error output is useful.  It's also possible that usermod should be ignoring this error.

One of our students was just hit by this at a Fedora 29 workstation.

He run:

$ sudo usermod -a -G dialout username

And got:

(Wed Dec 19 11:58:00:558187 2018) [sss_cache] [confdb_get_domains] (0x0010): No domains configured, fatal error!

Could not open available domains

usermod: sss_cache exited with status 2

usermod: Failed to flush the sssd cache.


I cannot reproduce it on my own machine or a virtual one.

I've changed the component to shadow-utils, so the maintainers are aware as well.

(In reply to Miro Hrončok from comment #2)
> One of our students was just hit by this at a Fedora 29 workstation.
> 
> He run:
> 
> $ sudo usermod -a -G dialout username
> 
> And got:
> 
> (Wed Dec 19 11:58:00:558187 2018) [sss_cache] [confdb_get_domains] (0x0010):
> No domains configured, fatal error!
> 
> Could not open available domains
> 
> usermod: sss_cache exited with status 2
> 
> usermod: Failed to flush the sssd cache.
> 
> 
> I cannot reproduce it on my own machine or a virtual one.
> 
> I've changed the component to shadow-utils, so the maintainers are aware as
> well.

shadow-utils only execs sss_cache, so sssd is the proper component. But for some reason (BZ upgrade?) I can't reassign the bug back.

Here you are, clicking the Component and hitting backspace worked for me.

Perhaps the sss_cache should be silent by default on such errors?

(In reply to Tomas Mraz from comment #5)
> Perhaps the sss_cache should be silent by default on such errors?

Yes and the errors should not be produced IMO (IOW, no domains should be just a no-op).

Also, I'm not sure how a no-domains error happens, there should always be at least the implicit files domain.

*** Bug 1661217 has been marked as a duplicate of this bug. ***

*** Bug 1662480 has been marked as a duplicate of this bug. ***

* master:
 * 415094687e92789060626176c5ced31d4122692d
 * 71475f1ed78a65d78f75e5ca0fdc6e20cfdf2f39
 * 325df4acae303efeabd96d2247fb5799c728536a
 * 88c0c3fcd1d97bd499bb28c2065ba19d629fa4f7

master:                                                                                                
* 159a2316b8d5560da5264022c598f1072f21bdba                                                              
* 2de3c5fb2490da0dabed0de498a8296db85a1e61

Has this been fixed? I'm not sure what the master and hash values represent? We still get this error and sssd is disabled.

(In reply to RobbieTheK from comment #11)
> Has this been fixed? I'm not sure what the master and hash values represent?

Upstream commits

> We still get this error and sssd is disabled.

Yes, but the patches were not backported to Fedora yet. Rather than backporting we'd like to release a new upstream version soon (Monday?) and include the fixes as part of a new version. There are several Fedora bugs that had been fixed in the meantime upstram and backporting everything would just be too tedious.

(In reply to RobbieTheK from comment #11)
> Has this been fixed? I'm not sure what the master and hash values represent?
> We still get this error and sssd is disabled.

If sssd is disabled then you can uninstall package sssd-common and you will not see any error.

*** Bug 1680446 has been marked as a duplicate of this bug. ***

(In reply to Miro Hrončok from comment #2)
> One of our students was just hit by this at a Fedora 29 workstation.
> 
> He run:
> 
> $ sudo usermod -a -G dialout username
> 
> And got:
> ...
> 
> I cannot reproduce it on my own machine or a virtual one.
> 
> I've changed the component to shadow-utils, so the maintainers are aware as
> well.

Not sure if it matters, but the machine I experienced the issue on originally had Fedora 22 or 24 or so. It has been through five or so system-upgrade.

I experienced the issue for the modem, too. The problem surfaced when tying to add myself to dialout group. The modem is new to the machine and was added last week for testing.

same problem with sss_cache in clamav 0.101.2

(In reply to customercare from comment #16)
> same problem with sss_cache in clamav 0.101.2

Please upgrade to sssd 2.1

https://bodhi.fedoraproject.org/updates/FEDORA-2019-efe7e2c5cf

We don't even use sssd and it's disabled. But clearing the.ldb files makes the message go away.


5.1.19-300.fc30
rpm -q sssd
sssd-2.2.0-3.fc30.x86_64

systemctl status sssd
● sssd.service - System Security Services Daemon
   Loaded: loaded (/usr/lib/systemd/system/sssd.service; disabled; vendor preset: enabled)
   Active: inactive (dead)


usermod -a -G cnslab ragarwal
(Sat Aug  3 18:30:02:247487 2019) [sss_cache] [sysdb_domain_cache_connect] (0x0010): DB version too old [0.20], expected [0.21] for domain implicit_files!
Higher version of database is expected!
In order to upgrade the database, you must run SSSD.
Removing cache files in /var/lib/sss/db should fix the issue, but note that removing cache files will also remove all of your cached credentials.
Could not open available domains
usermod: sss_cache exited with status 69
usermod: Failed to flush the sssd cache.
(Sat Aug  3 18:30:02:271662 2019) [sss_cache] [sysdb_domain_cache_connect] (0x0010): DB version too old [0.20], expected [0.21] for domain implicit_files!
Higher version of database is expected!
In order to upgrade the database, you must run SSSD.
Removing cache files in /var/lib/sss/db should fix the issue, but note that removing cache files will also remove all of your cached credentials.
Could not open available domains
usermod: sss_cache exited with status 69
usermod: Failed to flush the sssd cache.

(In reply to RobbieTheK from comment #19)
> We don't even use sssd and it's disabled. But clearing the.ldb files makes
> the message go away.
> 

The debug message says it need to upgrade version of sssd cache
and therefore it is a different issue.

But I have a little bit better solution for you If you do not want to use sssd at all.
Just remove package sssd-common and user{mod,add} will not be able to execute binary
sss_cache.

> 
> 5.1.19-300.fc30
> rpm -q sssd
> sssd-2.2.0-3.fc30.x86_64
> 
> systemctl status sssd
> ● sssd.service - System Security Services Daemon
>    Loaded: loaded (/usr/lib/systemd/system/sssd.service; disabled; vendor
> preset: enabled)
>    Active: inactive (dead)
> 
> 
> usermod -a -G cnslab ragarwal
> (Sat Aug  3 18:30:02:247487 2019) [sss_cache] [sysdb_domain_cache_connect]
> (0x0010): DB version too old [0.20], expected [0.21] for domain
> implicit_files!
> Higher version of database is expected!
> In order to upgrade the database, you must run SSSD.
> Removing cache files in /var/lib/sss/db should fix the issue, but note that
> removing cache files will also remove all of your cached credentials.
> Could not open available domains
> usermod: sss_cache exited with status 69
> usermod: Failed to flush the sssd cache.
> (Sat Aug  3 18:30:02:271662 2019) [sss_cache] [sysdb_domain_cache_connect]
> (0x0010): DB version too old [0.20], expected [0.21] for domain
> implicit_files!
> Higher version of database is expected!
> In order to upgrade the database, you must run SSSD.
> Removing cache files in /var/lib/sss/db should fix the issue, but note that
> removing cache files will also remove all of your cached credentials.
> Could not open available domains
> usermod: sss_cache exited with status 69
> usermod: Failed to flush the sssd cache.

I'm starting to wonder if shadow-utils shoudl just throw away all stderr from sss_cache on purpose and ignore the error code..

That could be done.

(In reply to Lukas Slebodnik from comment #20)
> (In reply to RobbieTheK from comment #19)
> > We don't even use sssd and it's disabled. But clearing the.ldb files makes
> > the message go away.
> > 
> 
> The debug message says it need to upgrade version of sssd cache
> and therefore it is a different issue.

Still seeing this in Fedora 31, should I open another bug?

groupadd: sss_cache exited with status 69
groupadd: Failed to flush the sssd cache.
(Mon Jan 13 11:46:15:017759 2020) [sss_cache] [sysdb_domain_cache_connect] (0x0010): DB version too old [0.20], expected [0.21] for domain implicit_files!
Higher version of database is expected!
In order to upgrade the database, you must run SSSD.
Removing cache files in /var/lib/sss/db should fix the issue, but note that removing cache files will also remove all of your cached credentials.
Could not open available domains
groupadd: sss_cache exited with status 69


> 
> But I have a little bit better solution for you If you do not want to use
> sssd at all.
> Just remove package sssd-common and user{mod,add} will not be able to
> execute binary
> sss_cache.

Did this, it removes a quite a few packages:

dnf remove sssd-common
Dependencies resolved.
===================================================================================================================================================
 Package                                  Architecture                  Version                              Repository                       Size
===================================================================================================================================================
Removing:
 sssd-common                              x86_64                        2.2.2-3.fc31                         @updates                        5.1 M
Removing dependent packages:
 sssd                                     x86_64                        2.2.2-3.fc31                         @updates                         34 k
 sssd-kcm                                 x86_64                        2.2.2-3.fc31                         @updates                        404 k
Removing unused dependencies:
 adcli                                    x86_64                        0.8.2-7.fc31                         @fedora                         268 k
 c-ares                                   x86_64                        1.15.0-4.fc31                        @fedora                         232 k
 cyrus-sasl-gssapi                        x86_64                        2.1.27-2.fc31                        @fedora                          45 k
 libdhash                                 x86_64                        0.5.0-43.fc31                        @fedora                          62 k
 libipa_hbac                              x86_64                        2.2.2-3.fc31                         @updates                         62 k
 libsss_autofs                            x86_64                        2.2.2-3.fc31                         @updates                         66 k
 libsss_certmap                           x86_64                        2.2.2-3.fc31                         @updates                        123 k
 libsss_idmap                             x86_64                        2.2.2-3.fc31                         @updates                         78 k
 libsss_nss_idmap                         x86_64                        2.2.2-3.fc31                         @updates                         89 k
 libsss_sudo                              x86_64                        2.2.2-3.fc31                         @updates                         59 k
 sssd-ad                                  x86_64                        2.2.2-3.fc31                         @updates                        387 k
 sssd-client                              x86_64                        2.2.2-3.fc31                         @updates                        258 k
 sssd-common-pac                          x86_64                        2.2.2-3.fc31                         @updates                        246 k
 sssd-ipa                                 x86_64                        2.2.2-3.fc31                         @updates                        717 k
 sssd-krb5                                x86_64                        2.2.2-3.fc31                         @updates                         78 k
 sssd-krb5-common                         x86_64                        2.2.2-3.fc31                         @updates                        286 k
 sssd-ldap                                x86_64                        2.2.2-3.fc31                         @updates                        165 k
 sssd-nfs-idmap                           x86_64                        2.2.2-3.fc31                         @updates                         42 k
 sssd-proxy                               x86_64                        2.2.2-3.fc31                         @updates                        144 k

Still on Fedora 33. sssd-2.4.1-1.fc33.x86_64

usermod -a -G ourgroup ouruser
(2021-02-14 20:39:35:810473): [sss_cache] [confdb_get_enabled_domain_list] (0x0040): Failed to get [domains] from [sssd], error [2] (No such file or directory)
(2021-02-14 20:39:35:810622): [sss_cache] [init_domains] (0x0020): Could not initialize domains
(2021-02-14 20:39:35:839051): [sss_cache] [confdb_get_enabled_domain_list] (0x0040): Failed to get [domains] from [sssd], error [2] (No such file or directory)
(2021-02-14 20:39:35:839127): [sss_cache] [init_domains] (0x0020): Could not initialize domains

I had to remove the above packages to bypass this error. Note sssd is not in use.

Appears to be an upstream bug as this user on Debian/Ubuntu reports it: https://piuparts.debian.org/sid/pass/sssd-tools_2.4.1-2.log and https://forum.ubuntu-it.org/viewtopic.php?p=5227042#p5227042

(In reply to RobbieTheK from comment #24)
> Still on Fedora 33. sssd-2.4.1-1.fc33.x86_64
> 
> usermod -a -G ourgroup ouruser
> (2021-02-14 20:39:35:810473): [sss_cache] [confdb_get_enabled_domain_list]
> (0x0040): Failed to get [domains] from [sssd], error [2] (No such file or
> directory)
> (2021-02-14 20:39:35:810622): [sss_cache] [init_domains] (0x0020): Could not
> initialize domains
> (2021-02-14 20:39:35:839051): [sss_cache] [confdb_get_enabled_domain_list]
> (0x0040): Failed to get [domains] from [sssd], error [2] (No such file or
> directory)
> (2021-02-14 20:39:35:839127): [sss_cache] [init_domains] (0x0020): Could not
> initialize domains
> 
> I had to remove the above packages to bypass this error. Note sssd is not in
> use.

Hi,

the reappearance of the messages is mostly related to the change of the default debug level which affects the command line tools as well, see e.g. https://github.com/SSSD/sssd/issues/5488.

bye,
Sumit