Rudder

When any information in LDAP about node is updated, the whole NodeInfo cache is invalidated - it's not a problem on small instances, or when there are no node properties, but on large systems with a lot of node properties, it can take more than 6 seconds, which locks virtually everything related to nodes.

Other symptoms are:
- LDAP using a lot of CPU,
- LDAP timeout in webapp logs.

We store the last modification date and all the ldap related objects, so a better strategy would be to:

1. check if there are entries with a newer modification date
2. if so, fetch all entries with the newer modification date
3. construct the new nodeinfo from these entries, completing with the cache for what's missing
4. replacing within the cache the updated entries