On 04/19/2016 07:46 PM, Fabrice Bacchella wrote:
Le 19 avr. 2016 à 17:35, Ondra Machacek <omachace@redhat.com> a écrit :
On 04/19/2016 04:37 PM, Fabrice Bacchella wrote:
I tried to plug ovirt using my company AD.
But I have a problem, the DNS srv records are not well managed and I can't use them so I changed pool.default.serverset.type from srvrecord to failover.
With AD you should use srvrecord, unless you have somehow miscofigured AD. Can you please elaborate more what does it mean 'DNS srv records are not well managed'?
The command dig +short _ldap._tcp.dsone.3ds.com any | wc -l return 122 lines. Out of that, I can only use less than 10, all other generates timeout. I don't know if it's firewall or forgotten DC that generate that. There is no way I can use srvrecord. This domain is totally out of my reach, I have to take it as is.
ok, that's not good, but if some of the domains which are working are in same site, you can use 'domain-conversion'(works only with srvrecord): pool.default.serverset.srvrecord.domain-conversion.type = regex pool.default.serverset.srvrecord.domain-conversion.regex.pattern = ^(?<domain>.*)$ pool.default.serverset.srvrecord.domain-conversion.regex.replacement = WORKING-SITE._sites.${domain}
Can you please send engine log or if you are on 3.6, then use this command to test and provide log: $ ovirt-engine-extensions-tool --log-level=FINEST --log-file=ad-search.log aaa search --entity-name=userX --extension-name=ad-authz
I kill it after 1h of execution, and a 1.6MB log file, when I have pool.default.serverset.type = srvrecord pool.default.serverset.srvrecord.domain = ${global:vars.domain}
With pool.default.serverset.type = failover and pool.default.connection-options.connectTimeoutMillis = 500, I got: time ovirt-engine-extensions-tool bla real 1m29.264s user 0m6.837s sys 0m0.291s and a 278KB log file.
And with my setup (pool.default.serverset.type and pool.default.dc-resolve.default.serverset.type set to failover, pool.default.connection-options.connectTimeoutMillis = 500), I got real 0m5.084s user 0m6.343s sys 0m0.164s and a 199KB log file.
With pool.default.dc-resolve.enable = false, the results is the same than with failover for every one.
Ok. So assure in your failover servers are GCs(for correct group resolution). Now it could use other servers (which you didn't specified in failover) in case you are resolving user/group from different domain, so it's chasing refferal, in that case we run 'dig domainX.forest.com A', so you can have actually more A records(inacessible) for it. Is that your case? Can you please share log of extensions-tool, so we can better understand your problem and provide better help.
Btw: Do you use mutli domain AD setup? Or only single domain?
I think it's a single domain, but I'm not a Microsoft expert at all.