Re: [Exim] Exim 3.16

Página superior
Eliminar este mensaje
Responder a este mensaje
Autor: Leonid Mamtchenkov
Fecha:  
A: Mangled
Cc: exim-users
Asunto: Re: [Exim] Exim 3.16
Hello Mangled,

Once you wrote about "[Exim] Exim 3.16":
M> Heres the problem i am having. I would much appreciate any help Thanx
M> :)
M>
M> About 8 months back i was using Redhat 6.0 with exim 2.10 and kernel
M> 2.2.9 and i had no problems, i then updated to Mandrake 7.1 kernel
M> 2.2.15 and exim 3.16, the problem i have is after say 25-30 days
M> i start getting the following output outputted to my screen and in my
M> logs as well.
M>
M> Mar 28 12:45:53 mail kernel: socket: no more sockets
M> Mar 28 12:45:53 mail kernel: grow_inodes: inode-max limit reached
M> Mar 28 12:45:53 mail kernel: socket: no more sockets
M> Mar 28 12:45:53 mail kernel: grow_inodes: inode-max limit reached
M> Mar 28 12:45:53 mail kernel: socket: no more sockets
M>
M> at this point exim stops delivering mail, and my whole system crashes
M> where i get permission denied errors and such in my logs, and also
M> permission denied errors or no such file or directory errors in my
M> exim logs, and i have no alternative but to reboot, and then after
M> reboot it is fine. until day 25-30 comes and it starts over again. at
M> first i thought it was linux / the kernel but i have 3 systems with
M> mandrake 7.1 , 2 of which runs exim 3.16 and the other does not, and
M> the one that does not is fine, so i came to the conclusion that it
M> must be exim, is there a way to rectify it? or should i increase my
M> inode-max limit ?, and if i say doubled the inode-max would then on
M> day 50 would i start getting this stuff again? , or is there something
M> else that i am missing?


That's not exim. That's your kernel talking. I had this problem beofre.
I sugggest you upgrade your kernel to 2.2.19. Meanwhile you might want to

echo NUMBER > /proc/sys/fs/inode-max

where NUMBER is some large number, which is greater then the one in
/proc/sys/fs/inode-nr . Try something like 125000 or 250000.

But that is defenetely is temporary solution, until your upgrade the kernel.

P.S.: you can also check the archives of the linux-kernel mailing list.

--
Best regards,
Leonid Mamtchenkov
E-mail: leonid@???
URL: http://www.leonid.maks.net