Open directory server crashing every 30 days / clients unable to connect to calendar, contacts server

I have uploaded an opendirectory log which contains some strange entries. This was about the time my clients were not able to connect anymore.

https://www.dropbox.com/s/a6gcwlhm3zbpkvc/opendirectoryd.log

Unfortunately, log files are a little bit beyond my comprehension. Is this log file useful or should I look for something else?

Thanks a lot!

DPSG-Scout

Message was edited by: DPSG-Scout

Hi Linc,

This looks the most relevant to me:

opendirectory.log

2014-03-11 11:13:09.460675 CET - 333.2628758.2628759 - Client: Python, UID: 93, EUID: 93, GID: 93, EGID: 93

2014-03-11 11:13:09.460675 CET - 333.2628758.2628759, Node: /Local/Default, Module: PlistFile - predicates with 'AND' are not supported

2014-03-11 12:09:00.296514 CET - State information (some requests have been active for extended period):

Sessions: {

28 -- opendirectoryd:

- {

Session ID: 7BFBA6FE-A968-4399-A129-E3A5945E2A81

Refs: singleton

Type: Default

Target: localhost

}

}

Nodes: {

43 -- authd:

- {

Node ID: 6D0E236D-6DBD-4E8C-BC01-B3F50C2C2D8E

Nodename: /LDAPv3/127.0.0.1

Session ID: <Default>

Refs: 1

Internal Use: X

an many more similar ones…

Thanks for your effort!

Thanks Linc. No, slapd.log is completely empty. The file was created in Dec 2013 and I had this issue twice since then.

Another guess?

ls -l /var/log/slapd.log

-rw-r----- 1 root admin 0 Dec 19 16:56 /private/var/log/slapd.log

Opened slapd.log with "sudo pico" to make sure it's not a permissions issue. Totally empty. Finder says zero bytes.

I am just thinking of some more info which might be useful: I am running the OS X Server installation inside a virtual machine in order to keep the system separated from the host (2012 quad core iMac) and have snapshots (doesn't help in this case though). Originally, I used Parallels Desktop, however, every month or so clients would not be able to connect anymore (just like now). They didn't even get a "wrong password" dialog. Instead, the Server.app would not open anymore and I had to restart OS X Server. After a reboot, the Server.app CPU stats showed a period with 0% CPU activity starting at the moment the server froze and ending with the reboot. I contributed this to Parallels Desktop 9 (maybe related to app nap, though disabling app nap didn't help) and migrated the entire VM to VMware fusion 6. Since then, the server wouldn't "freeze" anymore (no 0% CPU), yet once a month I am still having a similar issue as described in my initial post.

Well, OS X inside Parallels or VMware normally behaves just fine, so I don't think running OS X Server inside a VM should be a problem per se. I could image though that the OS X Server installation or opendirectory database is screwed up for whatever reason.

Do you think migrating calendars/contacts using the command line (suppose I can get this working with the workaround linked above), manually adding the users and adjusting the UUIDs afterwards could give me a working calendar server?

Or do you know of a better idea for starting with a clean installation or at least re-create the opendirectory database?

For now, I have taken the easiest way, created a clean OS X installation and used migration assistant to migrate all data. Let's see whether this will keep the server running for more than 30 days.

Interestingly, slapd.log has various entries now.