Hi, there were two responses. It is a bug in ncsd, for Solaris 8 the only solution for now is restarting nscd every now and then until a patch is available. Jan-Olaf Droese: same situation on some Solaris 2.6 boxes. Only stoping and restarting nscd helps. Joel Minsky (Sun): suggested to run nscd -g. I did this a day after restarting nscd and there were 25000 entries in the hosts cache (the box is a very busy web server), looked on sunsolve and found Bugs 4344735 and 4159699). Joels answer: Bingo, bug 4344735 is a duplicate of 4159699. There are patches for 2.6 and 7, I don't see one for Sol 8 but the fix is supposedly integrated into update 4, which I think is the 7/01 release. This looks like what you're seeing, nscd isn't flushing old entries, but instead creates a huge database. From bug 4159699: ... So restart nscd every once in a while, or get update 4, or open a call with sun to try and ask for a patch. original question was: > lately a box running Solaris 8 got slower and started paging with high io > rates, uptime was 110 days. The reason was the process nscd getting quite > big. top showed nscd consuming 150MB RES out of the 512MB memory > installed, so naturally there wasn't enough memory for other things. We > did not change /etc/nscd.conf, the box acts mainly as a loaded http and > mail server. After stopping and starting nscd everything was ok again. Are > there (still) problems with nscd? Thanks all, Michael -- Michael Hase Six Offene Systeme GmbH michael@six.de Sielminger Str. 63 http://www.six.de 70771 Leinfelden-Echterdingen phone +49 711 99091 62 GermanyReceived on Thu Aug 23 19:21:58 2001
This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:25:02 EDT