Hmmm,
No relies at all.
I checked sunsolve again, and finally found patch 105401-25
we had 105401-20 installed.
-- Problem Description:4165775 Replica servers may attempt to allocate huge amounts of memory
(from 105401-24)
4216240 nisd patch 105401-20 generates irrelevant messages "isn't a directory"
(from 105401-23)
4186012 NIS+ replicas may loose synchronization with the NIS+ master
(from 105401-22)
4055704 NIS_CALLBACK not authenticated (reworked) 4055715 NIS_PING not authenticated (reworked) 4055724 NIS_CHECKPOINT not authenticated (reworked) 4055727 NIS_CPTIME not authenticated (reworked)
(from 105401-21)
4055704 NIS_CALLBACK not authenticated 4055715 NIS_PING not authenticated 4055724 NIS_CHECKPOINT not authenticated 4055727 NIS_CPTIME not authenticated
(from 105401-20)
4162879 After the TTL value of directory cache expires the subdomain client. root object
(from 105401-19) --
Installing the uprev fixed the memory problem, and reduced our error logs considerably.
Still no suggestions about the NIS reorg'. We will have to set up a test domain and test a few ideas.
------------------------------------------------------------------------- Organization: Ford Motor Company of Australia Limited. ACN 004 116 223 Date: Tue, 14 Mar 2000 16:11:52 +1100 From: "Brian O'Connor. (CF583173) HO 2nd Floor" <boconno6@mailhub.foa.ford.com> To: sun-managers@sunmanagers.ececs.uc.edu Subject: A NIS+ question, and nisd problem X-Mailer: Mutt 1.0pre3i
Hello Sun Managers we wish to reorganise our NIS+ structure. At present we have a top domain and and two subdomains, eg
top / \ / \ sub1 sub2
we are considering changing the name of the masters of the two sub domains. At present the masters are both home directory file servers AND NIS+ masters. This makes replicas useless, since if the master fails there are no home dirs anyway.
I'm aware that it is quite involved to change the name of the root master, but what about the domain masters?
We are running 2.6(Generic_105181-15)
In addition we are finding that one of our replicas is grinding to a halt. rpc.nisd grows to > than 1000MB in a top listing and keyserv appears to die;
we get a lot of
nisd[613]: check_updaters: Unable to fork readonly process.
in the logs
a "showrev -p | cut -d" " -f1,2 | sort"
says that we have the following patches( I have deleted uprevs)
<snip>
Your opinions or help would be appreciated. -------------------------------------------------------------------------
Brian -- ------------------------------------------------------------------------- Brian O'Connor Ford Email: boconno6@ford.com Unix Consultant to Ford Telephone: +61 3 93597848 Ford Of Australia Ford Facsimile: +61 3 93598266 -------------------------------------------------------------------------
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:05 CDT