Michael Ströder wrote:
Howard Chu wrote:
John Lewis wrote:
What if I wanted to write a OpenLDAP backend for a systemd journal file or Elasticsearch so I can present my logs as an LDAP subtree so I can use my LDAP tools to filter my logs? Should I use back-shell for prototyping? If so, what is the usual work flow?
back-shell might work for rough prototyping. back-sock would be more reasonable these days.
For prototyping a back-sock listener in Python you could give module slapdsock a try:
https://pypi.python.org/pypi/slapdsock
Personally I use it for OATH-LDAP's bind listeners which seem to work fairly robust on moderate load. But the release 0.5.2 should also work with all other request types.
If you have a non-trivial deployment the sheer amount of log data can cause some interesting performance issues.
Indeed. Still it's an interesting idea; I've often thought about writing an ElasticSearch replacement on top of OpenLDAP. In a native backend it would be orders of magnitude faster than their stuff.
I'm glad both of you already knew where I was going with this.
Yes, I did want to write it in python Michael. Yes, I also think OpenLDAP is faster than ElasticSearch Howard.
On Wed, 2017-05-31 at 14:43 +0100, Howard Chu wrote:
Michael Ströder wrote:
Howard Chu wrote:
John Lewis wrote:
What if I wanted to write a OpenLDAP backend for a systemd journal file or Elasticsearch so I can present my logs as an LDAP subtree so I can use my LDAP tools to filter my logs? Should I use back-shell for prototyping? If so, what is the usual work flow?
back-shell might work for rough prototyping. back-sock would be more reasonable these days.
For prototyping a back-sock listener in Python you could give module slapdsock a try:
https://pypi.python.org/pypi/slapdsock
Personally I use it for OATH-LDAP's bind listeners which seem to work fairly robust on moderate load. But the release 0.5.2 should also work with all other request types.
If you have a non-trivial deployment the sheer amount of log data can cause some interesting performance issues.
Indeed. Still it's an interesting idea; I've often thought about writing an ElasticSearch replacement on top of OpenLDAP. In a native backend it would be orders of magnitude faster than their stuff.
John Lewis wrote:
I'm glad both of you already knew where I was going with this.
Yes, I did want to write it in python Michael.
If Howard et al have no objections you can ask questions here if you run into issues using the slapdsock module.
Ciao, Michael.
openldap-technical@openldap.org
-
Howard Chu -
John Lewis -
Michael Ströder