Re: Squeeze: sometimes, bind times out (backgrounded) at boot time
Ok, I really mixed things up. I'm sorry (and I'm also very sorry for
the *huge* delay in answering to this thread).
I meant that **ypbind** fails to bind to ypserver.
And yes, the NIS domain servers are specified in yp.conf by their
fully qualified names, and those names are hardcoded in /etc/hosts
file. Also, /etc/nsswitch.conf has hosts line as below:
hosts: files nis mdns4_minimal [NOTFOUND=return] dns mdns4
So, the ypbind should get the correct IPs for the servers immediately.
But what I see, in practice is: most times, the machine(s) gose up
properly. Other times, I see a timeout notice at boot
("...backgrounded"), and the system comes up unable to mount the
remote users' "home" directories. When that happens, normally
rebooting several times doesn't solve enything. Restarting nis and
autofs, in this order *does* solve the issue.
Best regards,
Joao Roscoe
PS. This is the second time I send this message, In the first time, I
got an weird automatic
response, something about "Case 80324" (googled for that, it's
somehing about a bug in php4 package).
Hope that was not me doing something really wrong. Double-checked the
"To" address content, just in case.
> On Thu, Sep 22, 2011 at 8:43 PM, Bob Proulx <bob@proulx.com> wrote:
>> Joao Roscoe wrote:
>>> I have a bunch of squeeze boxes running with nis and autofs. All are working
>>> well, no performance issues. However, at boot time, sporadically, bind times
>>> out, and the machine goes up without nis.
>>
>> Your words say "bind times out" and "nis" fails but what does bind
>> have to do with nis? When NIS/YP was written it was written for
>> systems that did not use BIND nor even have it installed. In a pure
>> NIS/YP system they used NIS/YP for host name resolution. NIS by
>> itself does not depend upon BIND. There isn't an intrinsic dependency
>> of one upon the other unless you have created one in your configuration.
>>
>>> Since home folders are NFS via autofs, the machine becames useless,
>>> and a reboot is required (I know that restarting nis and autofs,
>>> will solve it, but that requires root access).
>>
>> This reads to me that you have an NIS problem not a BIND problem.
>> Probably your BIND configuration is okay. Instead look for your
>> problem in your NIS configuration.
>>
>>> Is there any way to increase the timeout of bind at boot time?
>>
>> First find the root cause of the problem. It seems unlikely that it
>> is BIND.
>>
>> What do you have in your /etc/yp.conf file? Are you specifying to
>> find the nis server by broadcast, by IP address, or by server name?
>>
>> Note that the default Debian yp.conf file contains this following warning:
>>
>> # IMPORTANT: For the "ypserver", use IP addresses, or make sure that
>> # the host is in /etc/hosts. This file is only interpreted
>> # once, and if DNS isn't reachable yet the ypserver cannot
>> # be resolved and ypbind won't ever bind to the server.
>>
>> It seems likely to me that you have placed host names in that file but
>> failed to heed the warning and place the host names in your
>> /etc/hosts. But keeping host names in /etc/hosts isn't wonderful.
>> Neither is using IP addresses. I recommend avoiding names there and
>> using the broadcast protocol to find the nis servers.
>>
>> domain example.com broadcast
>>
>> That would allow a client to associate with any of the nis master and
>> slaves as they become available.
>>
>> Bob
>>
Reply to: