Description of problem:
It is still not clear to all users that running nscd and sssd at the same time for the same maps might not be the best idea because then there are two caching stacks that compete with each other.
Even though sssd already warns to syslog when it detects that nscd is running and the maps that nscd caches are the same that sssd caches, perhaps it would also help if nscd.conf had a comment telling something along those lines as well
Other ideas are welcome as well.
(In reply to Jakub Hrozek from comment #0)
> Description of problem:
> It is still not clear to all users that running nscd and sssd at the same
> time for the same maps might not be the best idea because then there are two
> caching stacks that compete with each other.
> Even though sssd already warns to syslog when it detects that nscd is
> running and the maps that nscd caches are the same that sssd caches, perhaps
> it would also help if nscd.conf had a comment telling something along those
> lines as well
Do you mean nsswitch.conf? I'm not sure if a warning makes sense in nscd.conf. The sss module is only listed in nsswitch.conf.
I assume he means this: https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html-single/system-level_authentication_guide/#usingnscd-sssd
I agree that mentioning this in /etc/nscd.conf is a good idea.
Talking about warnings: It might be also a good idea to mention the fact the "shared yes" disables the "cache hit rate" stats output of "nscd -g" (it'll show "0%").
Posted patches for Rawhide:
I added the following warning:
++# WARNING: Running nscd with a secondary caching service like sssd may lead to
++# unexpected behaviour, especially with how long entries are cached.
to both nscd.conf and nsswitch.conf.
I also added the following note to nscd.conf:
+ # shared <service> <yes|no>
++# NOTE: Setting 'shared' to a value of 'yes' will accelerate the lookup
++# with the help of the client, but these lookups will not be
++# counted as cache hits i.e. 'nscd -g' may show '0%'.
I think that covers all suggestions in this issue.
Looks good, thank you very much.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.