Dear sun-managers, regarding my YP question I got several responses, but none of them did help me with my problem. Since most of them should be helpful in other cases, I'll summarize them below anyway. In our case the problem turned out to be a remote YP slave-server which updates its maps every 30 minutes via a cron job. The leased line to that server has problems since yesterday (in one direction it has a large packet loss). The unsuccessful tries to transfer the maps seem to bog the master server down. Since stopping these jobs, my YP master is fine again. Seems as if the master ypserv gets locked by these transfers. Here are the other suggestions I received by mail: - check disk space in /var (we had plenty) - install the recent patches, especially 105403-04 (we did, also 108890-01 and 108893-01, didn't help. The only difference is that ypbind now writes "unable to set maximum RPC record size" upon startup, but otherwise works) - use ypwhich to determine to which master ypbind binds (we did this before posting. The master was bound to itself, as it should be. Things got better when we used ypset to bind the master's ypbind to one of the slave servers. So I knew the problem is ypserv, not ypbind) - check the usual 100Mbit/FDX/autonegotiation stuff for the network card (all is fine here. The network interface is not part of our problem since the ypserv and ypbind are on the same machine) - check that the maps are correct (they were, otherwise I would have had problems on the slaves as well) - run a script which automatically restarts the whole yp stuff, when the binding fails (ugh...) Thanks to Mark Gosselin, Nabeel Ahmed, Richard J. Buchman, John S. McNab, Bertrand Hutin, David Foster and Suresh Rajagopalan for their suggestions. Jan-Olaf Original Question: today we are observing some strange behaviour of ypbind on our YP master server: Every once in a while, ypbind logs "NIS server not responding for domain XXX" and everything which tries to resolve something via YP hangs. Sometimes it comes back with "NIS server for domain "XXX" OK" after a couple of minutes, sometimes it needs a 'ypstop;ypstart' to get things running again. ypserv and ypbind are running on the same machine, which is also the YP master for domain XXX. All other ypbinds which are bound to slave servers are fine. Does anyone have an idea what could cause this problem and what to do to fix it? I have one other machine where I had the same problem but since it occured only at night and only once a month I decided to ignore it. Both machines are running Solaris 2.6. -- Jan-Olaf Droese | Lahmeyer International GmbH email: zdjod@lif.de | Friedberger Strae 173 phone: +49-6101-55-1611 | 61118 Bad Vilbel fax: +49-6101-55-1623 | GermanyReceived on Fri Jun 29 15:34:17 2001
This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:24:58 EDT