sendmail 8.7.1 can't fork, not enough space

sendmail 8.7.1 can't fork, not enough space

Post by Sylvain Mart » Thu, 30 Nov 1995 04:00:00



I've been running sendmail 8.7 and 8.7.1 for a few weeks now and since I've
switched from 8.6.9, I'm getting tons of fork:not enough space. I've noticed
that it seems to happen whenever it tries to requeue unsent messages. I'm
running a Sparc 10MP with 128 MB of ram and over 600MB of swap. Everytime it
starts queueing, my swap goes down to 0MB but I still have about 30MB of main
memory free and I only have a mailq of only about 20 messages and no more. Any
ideas on what might be wrong?
 
 
 

sendmail 8.7.1 can't fork, not enough space

Post by Charles Stephe » Fri, 01 Dec 1995 04:00:00


 SM> I've been running sendmail 8.7 and 8.7.1 for a few weeks now and
 SM> since I've switched from 8.6.9, I'm getting tons of fork:not
 SM> enough space. I've noticed that it seems to happen whenever it
 SM> tries to requeue unsent messages. I'm running a Sparc 10MP with
 SM> 128 MB of ram and over 600MB of swap. Everytime it starts
 SM> queueing, my swap goes down to 0MB but I still have about 30MB of
 SM> main memory free and I only have a mailq of only about 20
 SM> messages and no more. Any ideas on what might be wrong?

Massive memory leak, film at 11.

Check to see if your sendmails are eating memory.  8.7.2 is out too,
so check it's README/Changes file and see if there was such a bug.
Upgrading to 8.7.2 might make it happy for no good reason as well.

Of course, it may be w/ the kernel.  Revert back to 8.6.9 and see if
you still have the problem.

cfs

 
 
 

1. ``Not enough memory'' -- 320 Megs not enough???

I'm getting baffled!

I have checked this out on three machines running 3.1.  Only the one
with 320 megs of memory bombs with ``Not enough memory'' while doing
this:

        mkhybrid -R -J -v -o /tmp/blah.iso /usr/src

I'm also able to make a 16 meg P60 fail, but she'll go way further
(mkhybrid of / -- does /usr/src plus lots more).

I've tried playing with ulimit, but that doesn't appear to make any
difference here.

I see swapctl -l always reporting 0 for used swap space.

To see what's happening, I've used the following:

        vmstat -w1
        ps -auxwk
        swapctl -l
        sysctl -a (just before and after bombing,
                   then diff the results)

I have yet to see anything obvious.  Does anybody have any ideas?
Anything?

Thanks!
Harry.

2. GMENU Segmentation Fault (Core Dump)

3. xfer [v]fork: not enough space

4. -= Linux Driving me Crazy =-

5. HELP -- fork: "not enough space"

6. POP accounts

7. fork: not enough space

8. Working ipfw for new kernel?

9. Solaris 2.7: "fork: not enough space"

10. Suninstall wth 'UPGRADE' not enough free space?

11. not enough space, but there is space

12. Apache problem: Not enough space: couldn't spawn child process

13. It's not bad canned meat...