You are not logged in.
After upgrading my system it no longer boot properly.
The kernel boots fine. Initrd executes and the normal init starts.
First of is the start of udev. This starts and immediately complains that nss_ldap is unable to contact the ldap server. Not very strange when networking isn't started yet.
After a little while and a few repetitions of the nss_ldap error message udev says its done.
After this activating swap fails but mounting Root readonly succeds. After this init proceeds with "Checking file system". This fails miserably saying "no such file or directory /dev/sda3" which is the previously mounted root filesystem.
It seems like udev gets busy when not able to connect to the ldap server and hence doesn't create the device nodes for my harddrive in time.
Because the check of the filesystem fails i end up on a rescue prompt. And when logging in to that and check /dev the sda* nodes are there.
Does anyone have a clue on how udev depends on nss_ldap? Any other thought on whats going on?
Offline
Ok, after some investigation it turns out that it seems like udev starts a few /sbin/vol_id on the disks/partitions attached. These processes in turn triggered nss_ldap which couldnt connect to the remote ldap server due to network not being up.
I have a less than beautiful solution to the problem. Altering nss_ldap to do soft binds instead of the default that is hard binding. When using hard bind and connect fails it uses an exponential backoff and tries again a little later. Changing to soft bind makes nss_ldap returning immediately when failing.
To change this edit /etc/nss_ldap.conf and change the line saying something like
#bind_policy hard
to
bind_policy soft
This at least helped me getting a system that boots again.
Offline