kNetwork Error

Question

kNetwork Error

I am regularly getting this error "kNetwork Error" when accessing my server via the Admin Tool, either locally on the machine or remotely with the app installed ay my workstation. Often I can just close the prompt and it goes away. Sometimes I close it and it locks me out of the server. I then have to remove it from the list, and add it back in. Mostly at it's .local address works, sometimes it only works at it's domain address.

We have a number of Xserves running 10.6.4

They are bound to our domain, and the directory services have been kerberized.

We share some share points via AFP and other with SMB. All our share points use AD for their groups.

I am also running DNS set to the loopback address in it's default configuration.

Those are all the services we're running. AFP, SMB, OpenDirectory (off but listed as part of the AD binding) and DNS.

I cannot get this error to go away. I have tried the system with and without DNS services. My concern is that other funky-ness we've experienced could be related to this. I'm not sure, I am somewhat skeptical. But to rule it out I'd like to resolve this issue.

I've searched but not found anything that says ah ha! I tried reading a link to a rather lengthy and thick tome on DNS. Again nothing there that screams out as an answer.

Anyone have any ideas on this?

45+ FCP Workstations using Decklink at work||an iMac and airbook at home, Mac OS X (10.5.6)

Posted on Nov 19, 2010 1:28 PM

Reply

Answer 1

Nov 19, 2010 1:32 PM in response to Martin Mcgowan

also, does anyone know if kerberos needs certain TCP/UDP ports open?

Reply

Answer 2

adm2p2l

Level 1

0 points

Nov 21, 2010 9:41 AM in response to Martin Mcgowan

same for me a few month ago with my first 10.6 xserve, dhcp service was disappeared and got this kNetwork Error when accessing DNS. Think thats a generic error, you have something wrong configured, DNS often, cos this is the first service you have to set up when you start OS X Server, you should start a clean install on this server
thats my idea

Reply

Answer 3

Nov 22, 2010 6:44 AM in response to adm2p2l

I've done a few clean installs and I'm getting this across the board.

The servers we have that are not bound/kerberized and are 10.5 don't have this issue.

Not sure what I'm doing wrong. I've tried setting the DNS a few ways.

Reply

Answer 4

adm2p2l

Level 1

0 points

Nov 23, 2010 9:57 PM in response to Martin Mcgowan

thats what i meant, you can not setting DNS a few ways, its a mess
did you upgrade this server from 10.5 to 10.6 ?
check your named.conf file and the zones folder

"In Mac OS X Server v10.6, the configuration and zone files used by Server
Admin have changed. If you edit the named.conf and zone files manually from
Terminal, the information is used by DNS. However, the information does not appear
in the DNS zones pane of Server Admin. Also, changes made in Server Admin are not
made to the named.conf file. It is recommended that you use Server Admin."

Reply

Answer 5

Nov 29, 2010 4:52 PM in response to Martin Mcgowan

Hey there, Martin

I'm having a similar issue. I've looked at other posts on this topic around the net (do a google search on: OS X Server Admin knetworkerror and you'll see 'em). Most of the posts I've found point to some sort of failure in bootpd. Check your system logs to see if you're noticing anything like that. I myself have not found a bootpd failure (as far as I know) but there are some curious failures happening in my logs:

From right around the time when the knetworkerror display pops up, I see the following in my system log:

11/29/10 4:14:19 PM servermgrd[9596] Exception in doCommand for module servermgr_dirserv: couldn't spawn task: path = /usr/sbin/slapconfig, errno -1
11/29/10 4:14:19 PM servermgrd[9596] --request was {
command = getState;
variant = withDetails;
}
11/29/10 4:21:50 PM Server Admin[494] Unexpected call to doMarkConfigurationAsDirty by 'DNS' plugin during updateConfigurationViewFromDescription

Also the following (note bootpd appears to restart right after this):

11/29/10 4:25:51 PM /usr/libexec/launchdadd[16430] FAILURE: Could not send MIG reply: 0x10000003
11/29/10 4:25:51 PM /usr/libexec/launchdadd[16430] FAILURE: Could not send MIG reply: 0x10000003
11/29/10 4:25:51 PM /usr/libexec/launchdadd[16430] FAILURE: Could not send MIG reply: 0x10000003
11/29/10 4:25:51 PM com.apple.launchd[1] (com.apple.servermgrd[9596]) Exited with exit code: 2
11/29/10 4:25:53 PM servermgrd[16436] servermgr_ipfilter:ipfw config:Notice:Flushed IPv4 rules
11/29/10 4:25:53 PM servermgrd[16436] servermgr_ipfilter:ipfw config:Notice:Flushed IPv6 rules
11/29/10 4:27:11 PM com.apple.launchd[1] (com.apple.suhelperd[16523]) Exited with exit code: 2
11/29/10 4:30:57 PM servermgrd[16436] [16436] error in getAndLockContext: flock returned -1 (errno=45) fd = 29 file was /private/var/servermgrd//servermgr_ipfilter.lock
11/29/10 4:31:02 PM bootpd[17650] server name mnemosyne.breitnet.net
11/29/10 4:31:02 PM bootpd[17650] interface en0: ip 10.23.0.7 mask 255.255.0.0
11/29/10 4:31:02 PM bootpd[17650] interface en1: ip 192.168.2.7 mask 255.255.255.0
11/29/10 4:31:02 PM bootpd[17650] bsdpd: re-reading configuration

Finally, the following appears in my system log over and over - I doubt it's connected but thought I'd include it anyway:

11/29/10 4:22:43 PM python[74] * -[NSAutoreleasePool release]: This pool has already been released, do not drain it (double release).
11/29/10 4:22:43 PM python[74] * -[NSAutoreleasePool release]: This pool has already been released, do not drain it (double release).
11/29/10 4:22:43 PM python[74] * -[NSAutoreleasePool release]: This pool has already been released, do not drain it (double release).

I'm running an Intel XServe RAID, 10.6.5, OD, AFP, SMB, DHCP, DNS, NetBoot, iCal, Web, Podcast Producer and XGrid (these last two hardly used at all, and mostly started as an experiment).

If anyone could shed light on this issue or any of the above errors it would be much appreciated!

Thanks!

Reply

Answer 6

Dec 1, 2010 11:53 AM in response to Picoscope

Saw some stuff related to bootpd out on the net but not in my system(s).

looked for errors in `/var/log/system.log/: and /Library/Logs/DiagnosticReports/bootpd2010-08-12-165459localhost.crash

I found nothing.

I verified my DNC Consistency per the article here -> http://support.apple.com/kb/ht3394

Still hunting. Of course this doesn't happen all the time. ti will clear out then come back.

Reply

Answer 7

Baguaman

Level 1

4 points

Jan 7, 2011 8:56 AM in response to Martin Mcgowan

I seem to have the "kNetwork Error" only when NFS and Netboot services are ON...
Fresh install on MAC PRO server 10.6.5 (AFP, SMB, DNS, OPEN D.,(NFS and NETBOOT))
Does any one else have the same experience?

Reply

Answer 8

BenDogAU

Level 1

0 points

Mar 22, 2011 11:54 PM in response to Martin Mcgowan

have you found a fix for this issue yet?

i'm having the same issue, i think it might be related to open directory as i can't seem to open that either.

after a few minutes of using Server Admin my machine will start throwing up kNetwork errors when i'm in Server Admin and i won't be able to change any settings until i reboot the server.

this is on a macmini server with latest OS

Reply

Answer 9

Mar 24, 2011 7:03 AM in response to BenDogAU

I have a MacMini Server 10.6 server install that I migrated from a 10.5 G5. Everything seems fine, but I keep getting an error stating that "The service has encountered an error" and asks me to refresh the view. It says in parenthesis kNetworkError. This appears on all the services we have running. I refresh and it appears fine.

From what little research I've done, many believe this to be a DNS issue, that perhaps there is something amiss in the DNS records. Unfortunately, we don't use this server to manage DNS, so we don't have that service running.

If anyone can shed some light on what a kNetworkError is, I would be very appreciative.

Reply

Answer 10

Mar 28, 2011 11:30 AM in response to Baguaman

Yes.

Reply

Answer 11

Jun 9, 2011 9:25 AM in response to Picoscope

Looking at my log, I see an error that appears to correlate to the failurs of the admin console:

6/9/11 11:07:22 AM servermgrd[19075] servermgr_ipfilter:ipfw config:Notice:Flushed IPv4 rules

6/9/11 11:07:22 AM servermgrd[19075] servermgr_ipfilter:ipfw config:Notice:Flushed IPv6 rules

6/9/11 11:36:15 AM servermgrd[19075] [19075] error in getAndLockContext: flock(servermgr_pcast) FATAL time out

6/9/11 11:36:15 AM servermgrd[19075] [19075] process will force-quit to avoid deadlock

I have a slight hing that this has to do with mDNS errors - it is the only other weird thing I see in my logs. Stay tuned...

Reply

Answer 12

Mark23

Level 4

994 points

Jun 9, 2011 12:14 PM in response to Martin Mcgowan

Martin Mcgowan wrote:

also, does anyone know if kerberos needs certain TCP/UDP ports open?

Set up the firewall service to block all ports except those listed here for directory, authentication, and administration protocols:

Open Directory Password Server uses ports 106 and 3659.
The Kerberos KDC uses TCP/UDP port 88, and TCP/UDP port 749 is used for Kerberos administration.
The shared LDAP directory uses TCP port 389 for an ordinary connection and TCP port 636 for an SSL connection.
When creating an Open Directory replica, keep port 22 open between the master and prospective replica. This port is used for SSH data transfer, which is used to transfer a complete, up-to-date copy of the LDAP database. After initial replica setup, only the LDAP port (389 or 636) is used for replication.
Workgroup Manager uses TCP port 311 and 625.
Server Admin uses TCP port 311.
SMB uses TCP/UDP ports 137, 138, 139, and 445.

Reply

Answer 13

binba

Level 1

24 points

Jul 21, 2011 6:03 PM in response to BenDogAU

BenDogAU wrote:

after a few minutes of using Server Admin my machine will start throwing up kNetwork errors when i'm in Server Admin

Same thing that I'm seeing, except that all I have to do to continue working normally is to restart Server Admin, not the server, so so far it's mostly been just an annoyance.

10.6.7 on XServe.

Reply

Answer 14

Oct 6, 2011 11:34 AM in response to binba

I'm finding that the Admin Tool doesnt' liek to be left open for long on the local machine. Mostly OK from my desktop.

Reply