Hi Yoann, Am 15.03.2013 um 18:49 schrieb Yoann Gini:
I’m Yoann. It’s my first message here so a little brief about me. I’m a OS X Server System Administrator and Trainer, actually working on a FreeBSD based setup for a simple services provider infrastructure.
I plan to use the calenderserver project for CalDAV and CardDAV instead of the easy to use SOGo to run a server more close as possible to the RFCs. You are not allone: I try to build a FreeBSD port for calendarserver, based on the 4.2 release.
I successfully setup the 4.2 release with a LDAP and a PostgreSQL backend on a FreeBSD 9.1 64 bits processor. I have (some how) working installations on 8.2 and 9.1, both 64 bit systems, using accounts.xml instead of LDAP.
It was a pain because the 4.2 didn’t handle well the 64 bits socket descriptors but I’ve add some fixes for that. I hit the same bug. It has been discussed here: http://www.mail-archive.com/calendarserver-dev@lists.macosforge.org/msg01611... That was the same with de dependencies, a lot of things seems to be unstable on 64bits, so I install the requirements by my own and run it with a -nr to don’t use built-in deps. My port does not use the run script at all, installing dependencies through the ports system.
By the way, the behavior to install all deps in .. fold is really unusual and put a big mess…
But what ever. It seems to look like a working config at the end.
I can run the server, authenticate my self, if I enable the principals navigation I can see all my users, I can connect iCal and use it. Well, it’s work. I just reached a similar point, where I start testing, using calendar and addressbook clients on OSX 10.6.8 and 10.8.3, but went soon in a problem, where the master process starts eating cpu cycles, as described here: http://www.mail-archive.com/calendarserver-dev@lists.macosforge.org/msg01618...
But, after some times, the server seems to by buggy, I start to make time outs for request, or for opening sockets, or sometime it rework again… Really random behavior and not at lot of interesting things in logs…
To have a idea, all my services, caldavd included are under nagios monitoring, I get something like 20 alerts per day about CalendarServer in a critical state…
Anyone can give me a clue about what to look ? What can be the source of the problem ? How to troubleshooting it ? I have no idea yet, but maybe you look through the above thread. Currently I'm trying to learn how to profile python processes...
Axel --- PGP-Key:29E99DD6 ☀ +49 151 2300 9283 ☀ computing @ chaos claudius