Forum Replies Created

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • in reply to: Import an LDIF file to Open Directory #368199
    elptacek
    Participant

    Way to late to be useful, I’m sure… try granting yourself a ticket:
    kinit root
    This seems to work for me.

    in reply to: TCP Listener returned error on accept #367910
    elptacek
    Participant

    The only thing besides OS X that runs on these servers is Retrospect, for backups.
    I have never resolved this issue. Instead I have installed a copy of the universal binary version of OS X server on my XServe G5s. I was able to do this using the same serial number that came with the PPC version of the OS. Apparently Apple doesn’t want people to do that, so if you try it, don’t blame me if it gets you in trouble.
    The good news is, this hasn’t happened since.
    The bad news is, my new XServe Xeons are randomly shutting themselves down.

    in reply to: TCP Listener returned error on accept #367715
    elptacek
    Participant

    I think I left this out of my original post — all of my servers are running 10.4.8. Most of my clients are running 10.4.7 and 10.4.8.
    Anyone know how to increase the listen queue depth for OS X? I am checking netstat and seeing listen queue overflows. IIRC, these are bad.

    –ep

    in reply to: Automount basics help? #367676
    elptacek
    Participant

    Try adding a line to /etc/syslog.conf on the server and the client for the daemon facility:
    daemon.* /var/log/daemon
    #> touch /var/log/daemon
    #> killall -HUP syslogd
    This should provide you with a wealth of logging information about what the automount daemon is doing. It’s not an answer, but it might help.
    ep

    in reply to: Problem wth AFP in golden triangle… #367675
    elptacek
    Participant

    Also check the groups permissions on the shares and the membership.
    One of the strangest things I’ve run aground of since upgrading our servers and OD master to 10.4.7 is that accounts that were migrated from netinfomanager (10.2) lack a generated uid field (you can see this in the Inspector tab in WGM). When I upgraded the OD master I was prompted to ‘upgrade legacy groups’ — I seriously wish I had not done this, because the action has caused a number of weird results for the older accounts that lack the generated uid. Best solution is to remove and restore the account.
    Another thing you can check is the path to the home directory. I have to manually edit the values for the NFSHomeDirectory attribute such that it reads /Users/Home/$USER rather than /Network/Servers/$SERVERNAME/Volumes/$PATH/$USER. The client’s automounter seems to require this.
    Hope this helps. 🙂

    –ep

    in reply to: TCP Listener returned error on accept #367620
    elptacek
    Participant

    I thought that the AFP being at maxThreads and hovering around 100% CPU use was the problem here, too. And usually when I begin seeing these messages, that is the case (also often when I’m not seeing them). Then it happened on one server when there were only 42 threads (instead of 600ish) and the %CPU was around 70. Kinda blew a hole in thread starvation theories.

    I’m eagerly (not really) awaiting the next incident so that I can run ktrace on the second AFS process before reboot.

    ep

    in reply to: FireFox/Mozilla and SMB homedirs #366942
    elptacek
    Participant

    Possibly useful: [url]https://wiki.inf.ed.ac.uk/DICE/OpenAFSUserIssueshere[/url]


    There is a conflict between the configuration file locking scheme used in Firefox and Thunderbird versions 1.5 and later, and the AFS locking implementation. This can result in Firefox and Thunderbird failing to start.
    A temporary workaround for this is to delete the .parentlock file that is in the configuration directory…
    * With Mozilla – rm -f ~/.mozilla/firefox/*.default/.parentlock
    * With Thunderbird – rm -f ~/.thunderbird/*.default/.parentlock

    in reply to: /LDAPv3/127.0.01 changed to /LDAPv3/10.70.1.135 #366805
    elptacek
    Participant

    This is what’s tripping me up. The reverse DNS lookup is good. OS X Server just seems to want /bin/hostname to return the FQDN.

    Curiouser and curiouser.

    in reply to: /LDAPv3/127.0.01 changed to /LDAPv3/10.70.1.135 #366801
    elptacek
    Participant

    It is not bound to itself, but the entry in Directory Access.app yesterday displayed the loopback and this morning it displayed /LDAPv3/10.70.1.135.

    I think I have found some clues:

    [code]
    host:~ root# changeip -checkhostname

    Primary address = 10.70.1.135

    Current HostName = host
    DNS HostName = host.domain.org

    To fix the hostname please run /usr/sbin/changeip for your system with the
    appropriate directory with the following values

    /usr/sbin/changeip <node> 10.70.1.135 10.70.1.135 host host.domain.org

    example:

    /usr/sbin/changeip /LDAPv3/127.0.0.1 10.70.1.135 10.70.1.135 host host.domain.org
    [/code]

    And this article: [url]http://docs.info.apple.com/article.html?artnum=303495[/url]

    So the mystery now is how the os installed sets the hostname to just host and OD appears to want it to be set to host.domain.org. There is no such thing as a reverse lookup that just returns the hostname without the domain appended, afaik.

    Also mysterious is how these mappings changed when I wasn’t looking (and probably asleep) but I have yet to find a log of exactly when and how this transpired.

    in reply to: /LDAPv3/127.0.01 changed to /LDAPv3/10.70.1.135 #366795
    elptacek
    Participant

    Boo. I posted in the wrong place.

    My apologies!

    Erin

Viewing 10 posts - 1 through 10 (of 10 total)