Hello all!
(Original Question at end)
Thanks to all, who replied. As ususal, this list is fast!
So far, I got 10 replies. Most of them pointed me to look at the
swap-space. Three ask, if we do a 'boot -r' after changing /etc/system
The solution was indeed a corrupt swap-partition.
Thanks again to:
Jamie Lawrence
Rick Flower
D'Arcey L. Carroll
Lee A. Fedor
Patricia Morgan
Marianne Rodgers
Christopher M. Conway
James E. Coby
Rahul Roy
Casper Dik
Answers in full text:
------------------------------------------
From: Jamie Lawrence <jal@ThirdAge.com>
I've seen this a couple of times when something ate all available
swap, so other processes can't fork.
Can't guarantee this is what is happening to you.
------------------------------------------
From: Rick Flower <Rick.Flower@trw.com>
Just a stupid comment... Did you halt the machine and do a "boot -r"?
I believe I read that in the Answerbook for updates to /etc/system...
------------------------------------------
From: "D'Arcey L. Carroll"
if they're workstations, use "top" to make sure that Xsun ( or any other
process ) isn't eating up all your swap space/memory. We always get this
message when active CDE users have had their computer up for a week or
so.
------------------------------------------
From: "Lee A. Fedor"
I have had this happen when I ran out swap space.
Try doing a swap -s and see if you have run out.
------------------------------------------
From: "Morgan, Patricia"
did you reboot after increasing parameters?
------------------------------------------
From: Marianne Rodgers
Did you try ulimit?
------------------------------------------
From: "Christopher M. Conway"
Not sure that this is your problem, but it's a good place to start.
There
are actually two process limits enforced. One is the overall number of
processes on the system, which you've fixed. There's also a per-user
limit
(maxuprc). You may need to increase that as well. The default value is
something like 50, I think. I always run it up to to at least 100, or
more
depending on the system and its typical use ( I routinely create 50-60
myself, so it's for my own convenience I do this.... 8^{)> ). The line
in
/etc/system looks like this:
set maxuprc=256
------------------------------------------
From: James E. Coby
Hello,
In regards to your no more processes message. Here is some info
I found to be interesting it is from :
" The Magic Garden Explained" by Good heart and Cox
ISBN 0-13-098138 -9
>From page 605 in chapter9 dealing with the use of the crash utility:
9.7.4 Memory Hints
In contrast to earlier releases, UNIX System V release 4 dynamically
allocates most of its data structures. This means that processes are
much more sensitive to memory allocation failure.
When a process fails to allocate dynamic memory, the error code EAGAIN
is returned.
When this erro is printed with the perror() library routine, it prints
the message "no more processes", which is misleading because this
error code applies to 3 seperate cases:
1. When a fork fails because the process table is full,
2. when the user is not allowed to create any more processes,
3. when a system call fails due to insufficient memory.
You may be able to see something using kmastsat from crash to
see if it is a memory problem. I'd also check with strstat to
see if it is a buffer problem.
Sorry I don't have any other suggestions but though you may be
interested
in this info.
------------------------------------------
From: Rahul Roy
Did you do a reconfiguration reboot after changing the /etc/system file
?
After making a change in the file - you have to do the following in
order
for the kernel to pick up and load the changes :
shutdown - to halt the machine
at the OK prompt - type "boot -r" ....
------------------------------------------
Original Question:
-----------------------------------------------------------------------
Hello all!
After upgrading our machines from 2.51 to 2.6 we get this error message:
No more processes
As described in various docs we have increased max_nprocs via the
maxuser variable:
Our /etc/system entry:
[...]
* We raise a couple of Kernel Parameters by raising maxusers.
*
* Kernel Parameter Description Default setting
*
* max_nprocs Number of Processes 10 + 16*maxusers
* ufs_ninode Number of Inodes max_nprocs + 16 +
maxusers + 64
* ncsize Dir. name lookup cache max_nprocs + 16 +
maxusers + 64
* ndquot Number of disk quotas maxusers*NMOUNT/4 +
max_nprocs
* maxuprc Number of Proc. per UID max_nprocs - 5
set maxusers=512
The problem: the error-message 'No more processes' still came.
Additional info:
ps -afe | grep wc
shows about 120 processes when this message came, not (10 +16*512) !
So, what is wrong?
Thanks in advance,
Stephan.
----------------------------------------------------------------------
------------------------------------------------------------------------ Fraunhofer-Institut fuer Software- und Systemtechnik (ISST) ---------------------------------------------------------------------- Stephan Grund | Telefon : 030/24306-376 System Support | Telefax : 030/24306-599 Mollstrasse 1 | E-Mail : Stephan.Grund@isst.fhg.de D-10178 Berlin | DFN-Mail : G=Stephan;S=Grund;O=ISST;P=FHG;A=D400;C=de
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:52 CDT