Solaris rpc.nisd taking lots o' memory

Solaris rpc.nisd taking lots o' memory

Post by Max D Hozv » Wed, 16 Nov 1994 14:29:54



Hi,

I am running Solaris 2.3 on Sparc LX, with 48 MB of memory.
I'm also running NIS+, and was wondering if anyone else has
noticed how much memory rpc.nisd seems to be taking:

# ps -lef | grep -v grep | grep PPID ; ps -lef | grep -v grep | grep rpc.nisd
 F S UID  PID PPID  C PRI NI   ADDR    SZ    WCHAN    STIME TTY     TIME COMD
 8 S root 143   1 80   1 20 fc371800  66111 f0178cf4   Nov 07 ?     2:33 /usr/sbin/rpc.nisd -r -Y                     ^^^^^                    
                                        |
                                        |
                                       big!

Is this correct for rpc.nisd (I checked Sunsolve for bug reports, and saw
nothing)?  ...and at this site, this box is the lone Sparc....so I have
nothing to compare it to.

-Max

 
 
 

Solaris rpc.nisd taking lots o' memory

Post by Casper H.S. D » Wed, 16 Nov 1994 18:25:03



Quote:>Hi,
>I am running Solaris 2.3 on Sparc LX, with 48 MB of memory.
>I'm also running NIS+, and was wondering if anyone else has
>noticed how much memory rpc.nisd seems to be taking:
># ps -lef | grep -v grep | grep PPID ; ps -lef | grep -v grep | grep rpc.nisd
> F S UID  PID PPID  C PRI NI   ADDR    SZ    WCHAN    STIME TTY     TIME COMD
> 8 S root 143   1 80   1 20 fc371800  66111 f0178cf4   Nov 07 ?     2:33 /usr/sbin/rpc.nisd -r -Y                     ^^^^^                        
>                                    |
>                                    |
>                                   big!

>Is this correct for rpc.nisd (I checked Sunsolve for bug reports, and saw
>nothing)?  ...and at this site, this box is the lone Sparc....so I have
>nothing to compare it to.

The solaris FAQ says:

4.8) Why is rpc.nisd such a memory pig according to ps?

    The good news is that it's not memory OR swap space you're
    being shown by 'ps'.  Instead it's showing you the process
    ADDRESS space which includes 256 MB of address space reserved
    for the NIS+ transaction log.  Given the cost of moving things
    around in memory and the fact that we have 4 GB of address
    space to play with it, this is a good idea.  You've just got to
    stop thinking small.  THINK BIG.  It's only 1/16th of the total
    process address space being used.  And if you ever exceed the
    256 MB size of the transaction log you're doing something VERY
    wrong.

    --- end of excerpt from the FAQ

Questions marked with a * or + have been changed or added since
the FAQ was last posted

The most recently posted version of the FAQ is available from
ftp.fwi.uva.nl in directory /pub/solaris

 
 
 

Solaris rpc.nisd taking lots o' memory

Post by Ray W. Hiltbran » Wed, 16 Nov 1994 23:16:32


Sparc 10 with 64megs of memory root master with memory SIZE being 3218
SS1000 with 256megs of memory root replica with memory size being 3085

What are the patch levels of your machine.  There have been a few NIS+
patches that have helped out...
        (from 101329-13)
          1155701 memory leak found in the NIS+ server code
          Memory leaks in the NIS+ code is using up all the system resource.

Kernal patch 101318-54 should fix the above NIS+ problem.


Engineering Network Services  |
Auburn University             |   http://www.eng.auburn.edu/~rayh/rayh.html
                              V
   If you are not in the lead your view never changes.


+Date:         Mon, 14 Nov 1994 21:29:54 -0800

+Subject:      Solaris rpc.nisd taking lots o' memory


+
+Hi,
+
+I am running Solaris 2.3 on Sparc LX, with 48 MB of memory.
+I'm also running NIS+, and was wondering if anyone else has
+noticed how much memory rpc.nisd seems to be taking:
+
+# ps -lef | grep -v grep | grep PPID ; ps -lef | grep -v grep | grep rpc.nisd
+ F S UID  PID PPID  C PRI NI   ADDR    SZ    WCHAN    STIME TTY     TIME COMD
+ 8 S root 143   1 80   1 20 fc371800  66111 f0178cf4   Nov 07 ?     2:33 /usr/sbin/rpc.nisd -r -Y                     ^^^^^
+                                        |
+                                        |
+                                       big!
+
+Is this correct for rpc.nisd (I checked Sunsolve for bug reports, and saw
+nothing)?  ...and at this site, this box is the lone Sparc....so I have
+nothing to compare it to.
+
+-Max

+

 
 
 

Solaris rpc.nisd taking lots o' memory

Post by William Mall » Fri, 18 Nov 1994 14:01:25



Quote:>Sparc 10 with 64megs of memory root master with memory SIZE being 3218
>SS1000 with 256megs of memory root replica with memory size being 3085

>What are the patch levels of your machine.  There have been a few NIS+
>patches that have helped out...
>        (from 101329-13)
>          1155701 memory leak found in the NIS+ server code
>          Memory leaks in the NIS+ code is using up all the system resource.

>Kernal patch 101318-54 should fix the above NIS+ problem.

No, no, no.  The kernel jumbo patch "fixes" it because "ps" has been modified
to only show the REAL size of rpc.nisd NOT the extra 256 Megabytes (65536 pages)
mmap()-ed for the Transaction Log.  Fixed in 2.4 as well.

rpc.nisd never had memory leaks THAT bad.  Not even in Solaris 2.0.

=wpm    William P. Malloy       SunSoft         Networking

 
 
 

1. rpc.nisd consuming lots of CPU

Hello!

I have a problem with rpc.nisd. We have about 20 Sparcstations running
Solaris 2.4, 5 of them run rpc.nisd.

Every morning, I find at least one of them, where rpc.nisd consumes a
lot of CPU power. I used truss to find out, it gets lots of requests
on fd5, and I used lsof, to find out, fd5 is an UDP socket. Then I use
snoop to find the origin ot these requests.

It is always one client station, and all requests fail with broken
credencials, something like broken seal. When I do a keylogout -f and
a keylogin on this machine, everything is fine.

Does anybody know, what goes wrong. I recently changed the secure RPC
passwords for every station, and I don't remember to have observed
this before. I did a keylogin -r on every client, and I checked,
secure RPC worked (at least I thing so).

It seems to work for some time, and suddenly break. I know, secure RPC
needs a good time, so I run xntp on all stations. On the servers, I
use the xntpd, on the clients I run ntpdate every hour from crontab.

Thanks for any help.

73, Mario
--

Institut fuer Robotik und Prozessinformatik der TU Braunschweig
Hamburger Strasse 267, 38114 Braunschweig, Germany

2. I can not compile postgresql. Help me!!

3. rpc.nisd memory size...

4. script to change an IP and or host name

5. Memory use by rpc.nisd on NIS+ replica

6. Knews

7. 2.3 rpc.nisd swallowing memory - is this normal?

8. Question: How does ipmasqadm work?

9. 5.6: NIS+: rpc.nisd -S0 won't allow nistbladm

10. Solaris 2.5.1 rpc.nisd looping

11. Solaris 2.3, rpc.nisd -S, DES or RSA ?

12. repeated failed open()'s results in lots of used memory [Was: [Fwd: memory consumption]]

13. rpc.nisd error message