000
@@ -297,6 +297,10 @@
extern u_long sb_max;
extern struct vm_zone *socket_zone;
extern so_gen_t so_gencnt;
+extern int maxripcb;
+extern int maxdivcb;
+extern int maxtcpcb;
+extern int maxudpcb;
struct file;
struct filedesc;
-Original Message-
From: Mike S
On Thu, Jul 17, 2003 at 08:45:29PM -0400, Scot Loach wrote:
> Bosko:
>
> The problem I have is that each of the four pcb types has its own zone that
> is preallocated to hold maxsockets pcbs. This is a waste of kva that could
> be better used. Since I'm not using divert sockets or raw sockets,
, and increase the others
as much as possible.
I should have a patch for this next week sometime.
-Original Message-
From: Bosko Milekic [mailto:[EMAIL PROTECTED]
Sent: Wednesday, July 16, 2003 5:52 AM
To: Scot Loach
Cc: '[EMAIL PROTECTED]'
Subject: Re: Kernel tuning for large maxsockets
On Tue, Jul 15, 2003 at 08:51:12PM -0400, Scot Loach wrote:
> True, I can add a tunable for each of tcp, udp, raw, divert.
>
> What will happen when the system runs out of pcbs?
You'll no longer be able to allocate and attach pcbs to sockets until
one of the current pcb consumers returns tha
On Tue, Jul 15, 2003 at 05:22:55PM -0400, Scot Loach wrote:
> Currently, whenever maxsockets is increased, this causes kernel memory to be
> preallocated for each type of pcb (tcp, udp, raw, divert). The number of
> pcbs preallocated for each of these is always the same as maxsockets.
>
> This i
On Tue, 15 Jul 2003, Scot Loach wrote:
> True, I can add a tunable for each of tcp, udp, raw, divert.
That's probably a good idea, I'm sure I'll end up messing with it a bit
once you send it to me. :)
> What will happen when the system runs out of pcbs?
I think it should handle everything fine
: Kernel tuning for large maxsockets
On 15 Jul, Mike Silbersack wrote:
>
> On Tue, 15 Jul 2003, Scot Loach wrote:
>
>> Is there any reason I should not modify the kernel code to only let a
small,
>> fixed number of raw and divert pcbs be preallocated instead of having
them
>
On 15 Jul, Mike Silbersack wrote:
>
> On Tue, 15 Jul 2003, Scot Loach wrote:
>
>> Is there any reason I should not modify the kernel code to only let a small,
>> fixed number of raw and divert pcbs be preallocated instead of having them
>> scale with maxsockets?
>
> Your idea is sound.
>
>> Nex
On Tue, 15 Jul 2003, Scot Loach wrote:
> Is there any reason I should not modify the kernel code to only let a small,
> fixed number of raw and divert pcbs be preallocated instead of having them
> scale with maxsockets?
Your idea is sound.
> Next, does this seem like a generally useful thing th
Currently, whenever maxsockets is increased, this causes kernel memory to be
preallocated for each type of pcb (tcp, udp, raw, divert). The number of
pcbs preallocated for each of these is always the same as maxsockets.
This is probably a waste of memory for raw sockets and divert sockets, since
10 matches
Mail list logo
