Peter Hessler | 1 Oct 01:06 2003

Re: linux isakmpd with freeswan - ADD: No buffer space available

Are you trolling?  This is a list for OpenBSD, not Linux.

BTW, check memory or hard drive usage.

On Tue, Sep 30, 2003 at 03:59:06PM -0700, Don Mahurin wrote:
:hi,
:
:I am trying to get isakmpd and Linux 2.6test6 to work with freeswan on another 
:box. I think I am close, as freeswan reports getting trough all stages:
:
:: 104 "don192" #68115: STATE_MAIN_I1: initiate
:: 106 "don192" #68115: STATE_MAIN_I2: sent MI2, expecting MR2
:: 108 "don192" #68115: STATE_MAIN_I3: sent MI3, expecting MR3
:: 003 "don192" #68115: ignoring informational payload, type
:: IPSEC_INITIAL_CONTACT
:: 004 "don192" #68115: STATE_MAIN_I4: ISAKMP SA established
:: 112 "don192" #68116: STATE_QUICK_I1: initiate
:: 004 "don192" #68116: STATE_QUICK_I2: sent QI2, IPsec SA established
:
:But, I cannot ping either way. And isakmpd seems to be releasing the SA. Any 
:thoughts on why this may be?
:
:102614.946707 Default pf_key_v2_set_spi: ADD: No buffer space available
:102614.947269 Timr 10 timer_remove_event: removing event 
:exchange_free_aux(0x8296e58)
:102614.947748 Mesg 20 message_free: freeing 0x8297270
:102644.688511 Sdep 20 pf_key_v2_expire: HARD dst 198.144.198.215 SPI b400ac31 
:sproto 3
:102644.692829 Timr 10 timer_remove_event: removing event 
:sa_hard_expire(0x8296f38)
(Continue reading)

Markus Friedl | 1 Oct 10:19 2003
Picon

Re: linux isakmpd with freeswan - ADD: No buffer space available

> Also, when compiling isakmpd, I had to define AES_BLOCK_SIZE to 16 to get it 
> to compile. Where should this be defined?
> 
> In file included from crypto.c:39:
> crypto.h:103: `AES_BLOCK_SIZE' undeclared here (not in a function)
> crypto.h:104: `AES_BLOCK_SIZE' undeclared here (not in a function)

AES should be probably protected by #if openssl >= 0.9.7

just remove the AES define from the features list in the makefile.

Magnus Larsson | 1 Oct 22:43 2003
Picon

Create an ramdrive???

Hi All.
I've tried to find out how to create an ramdrive within normal 
operation. No boot cd with ramdrive.. But I have'nt been able to find it.
Please help. How do I create it?
Best regards Magnus
--

-- 
      _    _
     (o)--(o)
    /.______.\
    \________/
   ./        \.
  ( .        , )
   \ \_\\//_/ /
    ~~  ~~  ~~

Have fun!!!
**** Magnus ****

Chuck Yerkes | 1 Oct 22:47 2003

Re: Create an ramdrive???

You'd ask on misc <at> 

or "man mfs"

Quoting Magnus Larsson (knatteman <at> telia.com):
> Hi All.
> I've tried to find out how to create an ramdrive within normal 
> operation. No boot cd with ramdrive.. But I have'nt been able to find it.
> Please help. How do I create it?
> Best regards Magnus
> -- 
>      _    _
>     (o)--(o)
>    /.______.\
>    \________/
>   ./        \.
>  ( .        , )
>   \ \_\\//_/ /
>    ~~  ~~  ~~
> 
> Have fun!!!
> **** Magnus ****

Alexandre | 5 Oct 19:21 2003
Picon

No buffer space available

Hi,

I'm having trouble with qos and openbsd3.3, basically I do ACKs 
prioritizing (www.benzedrine.cx/ackpri.html) and when there is a lot of 
outbound traffic (128kbps is the max on my adsl link), the network 
traffic stops after a couple of minutes, except already established 
connections. If I do a ping I get this message "No buffer space available".

I've stfa and there seem to be other people with same problem, so as 
suggested in other posts I've tried different NICs and increased network 
buffers via sysctl but it does not make any difference.

When the problem occurs, if I reload the pf rules with pfctl -f 
/etc/pf.conf if works again for a couple of minutes. If I remove qos 
from the rules, there is no problem, so it has definetely something to 
do with qos!

Any ideas ?

Regards,
Alexandre

Todd C. Miller | 5 Oct 21:29 2003

Re: No buffer space available

Getting ENOBUFS is intended behavior--it is how altq indicates
to userland that the queue is full.

It sounds like your queue rules in pf.conf are the problem.  You
have to be very careful when prioritizing things that you don't
lock out important stuff (like outbound ICMP or DNS for example).
You probably want several priority levels.

 - todd

Alexandre | 6 Oct 19:07 2003
Picon

Re: No buffer space available

That has solved the problem, looks like my qos rules were too extreme :)
Thanks.

Todd C. Miller wrote:

> Getting ENOBUFS is intended behavior--it is how altq indicates
> to userland that the queue is full.
> 
> It sounds like your queue rules in pf.conf are the problem.  You
> have to be very careful when prioritizing things that you don't
> lock out important stuff (like outbound ICMP or DNS for example).
> You probably want several priority levels.
> 
>  - todd

Giannici Federico | 7 Oct 10:26 2003
Picon

Panic: ffs_alloccg: map corrupted

My system paniced a couple of times in a couple of days with the 
following error: "ffs_alloccg: map corrupted"!

Furthermore, I have noticed some strange things occur simetimes: a 
particular byte of a give file changes!

The system is an OpenBSD 3.4 i386 (a checkout of OPENBSD_3_4 of a couple 
of days ago).

What does it means?

Thanks.

P.S.
Here it is the dmesg:

Oct  7 10:15:54 galadriel /bsd: 
giannici <at> galadriel.neomedia.it:/usr/src/sys/arch/i386/compile/NEOMEDIA
Oct  7 10:15:54 galadriel /bsd: cpu0: Intel Pentium 4 ("GenuineIntel" 
686-class) 2.81 GHz
Oct  7 10:15:54 galadriel /bsd: cpu0: 
FPU,V86,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CFLUSH,ACPI,MMX,FXSR,SIMD,SIMD2,SS,HTT,TM,SBF,CNXT-ID
Oct  7 10:15:54 galadriel /bsd: real mem  = 1072476160 (1047340K)
Oct  7 10:15:54 galadriel /bsd: avail mem = 988934144 (965756K)
Oct  7 10:15:54 galadriel /bsd: using 4278 buffers containing 53727232 
bytes (52468K) of memory
Oct  7 10:15:54 galadriel /bsd: mainbus0 (root)
Oct  7 10:15:54 galadriel /bsd: bios0 at mainbus0: AT/286+(00) BIOS, 
date 06/11/03, BIOS32 rev. 0  <at>  0xf0010
Oct  7 10:15:54 galadriel /bsd: pcibios0 at bios0: rev. 2.1  <at>  
(Continue reading)

Dave Hines | 7 Oct 11:22 2003

Re: Panic: ffs_alloccg: map corrupted

Hmmm - it may be worth running a memory test for a day or so, see
http://www.memtest86.com/

(I usually make this a boot option on my x86 machines, & run it if
anything unexpected starts happening).

Dave.

On Tue, Oct 07, 2003 at 10:26:37AM +0200, Giannici Federico wrote:
> My system paniced a couple of times in a couple of days with the
> following error: "ffs_alloccg: map corrupted"!
>
> Furthermore, I have noticed some strange things occur simetimes: a
> particular byte of a give file changes!
>
> The system is an OpenBSD 3.4 i386 (a checkout of OPENBSD_3_4 of a couple
> of days ago).
>
> What does it means?
>
> Thanks.

Giannici Federico | 7 Oct 15:38 2003
Picon

Re: Panic: ffs_alloccg: map corrupted

Dave Hines wrote:
> Hmmm - it may be worth running a memory test for a day or so, see
> http://www.memtest86.com/

I have run it for a while but no error was found...

If nobody else have problems with FFS under 3.4, then could it be a 
problem with the gdt driver (software or hardware)?

Thanks.

> On Tue, Oct 07, 2003 at 10:26:37AM +0200, Giannici Federico wrote:
> 
>>My system paniced a couple of times in a couple of days with the
>>following error: "ffs_alloccg: map corrupted"!
>>
>>Furthermore, I have noticed some strange things occur simetimes: a
>>particular byte of a give file changes!
>>
>>The system is an OpenBSD 3.4 i386 (a checkout of OPENBSD_3_4 of a couple
>>of days ago).
>>
>>What does it means?
>>
>>Thanks.

--

-- 
___________________________________________________
     __
    |-                      giannici <at> neomedia.it
(Continue reading)


Gmane