On Sun, 2018-07-15 at 09:48 -0700, Matt Weatherford wrote: > I went to upgrade my debian 9 system today and got a prompt at the > nslcd upgrade to enter the uri path. I entered this and the upgrade > hung. I am running upgrades over ssh sessions in non-interactive > mode and wonder if there is something about this upgrade that is > breaking the script nslcd is running.
Just to be clear: this problem happened when upgrading from 0.9.7-2 to 0.9.7-2+deb9u1? I recently fixed an issue related to the nslcd start-up that could result in a delay of 1 minute if nscd is used (will be fixed in 0.9.10). Another issue I have seen is that restarting slapd can take a long time. Depending on how your configuration was done normal upgrades of nslcd should not prompt for configuration and there shouldn't have been any major changes to this since around 0.9.1 (nothing major in the packaging between jessie and stretch). > It seems the new script is parsing the nslcd.conf file and making > changes and / or normalizing things. Is there a changelog that I can > review to better understand what is happening here? The whole debconf configuration is a bit tricky because we need to support debconf pre-seeding, load configuration from an existing configuration file and we have some magic to detect sane defaults if no pre-seeding was done and no existing configuration is in place. > What additional info would be helpful to you that I can gather? If you can reproduce the problem a log of the upgrade and where it was hanging would be useful. > Versions of packages nslcd suggests: > pn kstart <none> > > -- debconf information: > nslcd/ldap-auth-type: SASL > nslcd/ldap-sasl-krb5-ccname: /var/run/nslcd/nslcd.tkt > nslcd/ldap-sasl-mech: EXTERNAL If you are using Kerberos authentication for the connection to the LDAP server you may want to use kstart to keep the ticket fresh. Thanks, -- -- arthur - [email protected] - https://people.debian.org/~adejong --
signature.asc
Description: This is a digitally signed message part
