Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-23 Thread Dag-Erling Smørgrav
Ganbold [EMAIL PROTECTED] writes:
 We are using several squid machines (6 machines, each have all others
 as a siblings) for transparent caching/proxying using gre tunnel and
 wccp2 (with Cisco router).  Can varnish work in such situation?

Probably not; Varnish is a reverse proxy.

DES
-- 
Dag-Erling Smørgrav - [EMAIL PROTECTED]
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-23 Thread N. Harrington
--- JoaoBR [EMAIL PROTECTED] wrote:
 On Tuesday 21 August 2007 20:54:36 N. Harrington wrote:
  Hello
   I feel stupid, but I am confused about kern.maxdsiz (or datasize via
  limits command) on FreeBSD amd64.
 
 
   I have seen many posts and suggestions to raise it to 1G. However it seems
  this only applies to i386. By default, on servers I have with 4G of
  physical memory, and 2X that of swap, I am showing a reported datasize
  limit of 33554432KB. far in excess of even my physical and swap combined! I
  have seen suggestions from mysql for maxdsize to be set to 1G. Obviously no
  such problem with amd64?
 
 on amd64 when maxdsiz not set it stops at 512 limit but soon you set it to 1G 
  
 it appearently is able to use more  this is not the case on i386 - but I 
 would say don't worry about it
 
 on i386 your machine could hang at boot when setting maxdsiz higher than 
 installed physical memory but that never happened to me with amd64
 
 I have some server running squid for caching perfect with 4, 8 and 16G of RAM 
 
 I set maxdsiz do 3G on machines with 4Gigs of RAM but I do not run anything 
 else so then I adjust cache_mem with maximum_object_size_in_memory to use the 
 most possible amount of memory without doing swap. 
 
 anyway you set it in boot.loader there is no need to compile something into 
 the kernel
 
 -- 
 
 João
 Thanks everyone.

  I tried setting my maxd size to 3.5G on a machine with 4gigs ram. It caused 
squid to seem to rop
out on occasion. 
 It seems so odd that on i366, maxd size is so small that one likely needs to 
set it higher to
allow access to more memory. However on amd64, is such a high number that one 
would need to lower
it to prevent accessing too much memory?  Something really odd about that.

 I also found that I could even double the datasize / maxd size to 2X 
33554432kB and it would boot
and run just fine on a machine with 8 gigs of ram. How weird is that! Did amd64 
just cause this
setting to add 3 zero's? I can see the legacy documentation for i386 BSD and 
applications is going
to cause some weirdness and problems if not careful. 



 As for squid, just like Sendmail and Apache, yes there and plenty of I do it 
better and I am
newer, alternatives out. But some things stay old favorites for a reason. I 
will take a look at
Varnish, but it seemed much less user friendly to me that squid and much much 
less feature laden.
Also, I saw no way to purge individual files from storage. (something I have to 
have) So it's
always nicer to know how to make things work with what one has or needs to use, 
rather than just
being told to use something else.

 I have over 30 machines with various configurations of squid in accelerator 
mode and most seem to
work fine. However I will say they do have a preference for running on (1) 
Single core cpu and
SCSI hard drives. On an average server with ~300G of disk, I have over 10M 
objects in storage. As
usual though, I am dealing with a program(squid) that seems to be, Linux first, 
it happens to also
run on FreeBSD second. Even though it seems many people seem to be using it 
with FreeBSD. 

 Squid has also been great for me to test/beatup on gjournal (which should be 
in 6 by now and be
available standard) and zfs.


 Nicole





The Large Print Giveth And The Small Print Taketh Away
 -- Anon
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Dag-Erling Smørgrav
Chuck Swiger [EMAIL PROTECTED] writes:
 You should configure squid to use no more than about 60 - 70% of the
 available physical RAM-- ie, set the cache_mem parameter to about 2.5
 or 3GB.

Better yet, don't run Squid at all.  It was designed for a computer
architecture that was already obsolete when Squid was first written.

 It wouldn't be unreasonable to limit datasize to 3 GB on such a
 machine, assuming that nothing you run will ever need to grow
 larger...

...actually, maxdsiz is meaningless in FreeBSD 7, because the new
allocator uses mmap(2) instead of brk(2) / sbrk(2), so malloc() counts
towards the resident set size (ulimit -m), not the data segment size
(ulimit -d).

(unless, of course, your application has its own allocator, in which
case you can kiss performance goodbye)

DES
-- 
Dag-Erling Smørgrav - [EMAIL PROTECTED]
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Ganbold

Dag-Erling Smørgrav wrote:

Chuck Swiger [EMAIL PROTECTED] writes:
  

You should configure squid to use no more than about 60 - 70% of the
available physical RAM-- ie, set the cache_mem parameter to about 2.5
or 3GB.



Better yet, don't run Squid at all.


Ok, then what do you recommend instead of Squid?

thanks,

Ganbold



  It was designed for a computer
architecture that was already obsolete when Squid was first written.

  

It wouldn't be unreasonable to limit datasize to 3 GB on such a
machine, assuming that nothing you run will ever need to grow
larger...



..actually, maxdsiz is meaningless in FreeBSD 7, because the new
allocator uses mmap(2) instead of brk(2) / sbrk(2), so malloc() counts
towards the resident set size (ulimit -m), not the data segment size
(ulimit -d).

(unless, of course, your application has its own allocator, in which
case you can kiss performance goodbye)

DES
  



--
Heuristics are bug ridden by definition. If they didn't have bugs, then 
they'd be algorithms.

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Claus Guttesen
  You should configure squid to use no more than about 60 - 70% of the
  available physical RAM-- ie, set the cache_mem parameter to about 2.5
  or 3GB.
 
 
  Better yet, don't run Squid at all.

 Ok, then what do you recommend instead of Squid?

Varnish.

http://varnish.projects.linpro.no/

-- 
regards
Claus

When lenity and cruelty play for a kingdom,
the gentlest gamester is the soonest winner.

Shakespeare
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Dag-Erling Smørgrav
Ganbold [EMAIL PROTECTED] writes:
 Dag-Erling Smørgrav [EMAIL PROTECTED] writes:
  Better yet, don't run Squid at all.
 Ok, then what do you recommend instead of Squid?

That depends on what you use it for...

DES
-- 
Dag-Erling Smørgrav - [EMAIL PROTECTED]
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Dag-Erling Smørgrav
Alexandre Biancalana [EMAIL PROTECTED] writes:
 Dag-Erling Smørgrav [EMAIL PROTECTED] writes:
  Ganbold [EMAIL PROTECTED] writes:
   Ok, then what do you recommend instead of Squid?
  That depends on what you use it for...
 What the options for forward proxy/cache with user authentication and
 access control ?

I'm not really into forward proxies, but there are a few (mod_proxy,
privoxy, wwwoffle).

DES
-- 
Dag-Erling Smørgrav - [EMAIL PROTECTED]
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Alexandre Biancalana
On 8/22/07, Dag-Erling Smørgrav [EMAIL PROTECTED] wrote:

 Ganbold [EMAIL PROTECTED] writes:
  Dag-Erling Smørgrav [EMAIL PROTECTED] writes:
   Better yet, don't run Squid at all.
  Ok, then what do you recommend instead of Squid?

 That depends on what you use it for...


What the options for forward proxy/cache with user authentication and access
control ?
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread JoaoBR
On Tuesday 21 August 2007 20:54:36 N. Harrington wrote:
 Hello
  I feel stupid, but I am confused about kern.maxdsiz (or datasize via
 limits command) on FreeBSD amd64.


  I have seen many posts and suggestions to raise it to 1G. However it seems
 this only applies to i386. By default, on servers I have with 4G of
 physical memory, and 2X that of swap, I am showing a reported datasize
 limit of 33554432KB. far in excess of even my physical and swap combined! I
 have seen suggestions from mysql for maxdsize to be set to 1G. Obviously no
 such problem with amd64?

on amd64 when maxdsiz not set it stops at 512 limit but soon you set it to 1G  
it appearently is able to use more  this is not the case on i386 - but I 
would say don't worry about it

on i386 your machine could hang at boot when setting maxdsiz higher than 
installed physical memory but that never happened to me with amd64

I have some server running squid for caching perfect with 4, 8 and 16G of RAM 

I set maxdsiz do 3G on machines with 4Gigs of RAM but I do not run anything 
else so then I adjust cache_mem with maximum_object_size_in_memory to use the 
most possible amount of memory without doing swap. 

anyway you set it in boot.loader there is no need to compile something into 
the kernel

-- 

João







A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura.
Service fornecido pelo Datacenter Matik  https://datacenter.matik.com.br
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread JoaoBR
On Wednesday 22 August 2007 08:32:05 Claus Guttesen wrote:
   You should configure squid to use no more than about 60 - 70% of the
   available physical RAM-- ie, set the cache_mem parameter to about 2.5
   or 3GB.
  
   Better yet, don't run Squid at all.
 
  Ok, then what do you recommend instead of Squid?

 Varnish.

 http://varnish.projects.linpro.no/

that certainly is a really hotty for caching :)


-- 

João







A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura.
Service fornecido pelo Datacenter Matik  https://datacenter.matik.com.br
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Chuck Swiger

On Aug 22, 2007, at 3:53 AM, Dag-Erling Smørgrav wrote:

Chuck Swiger [EMAIL PROTECTED] writes:

You should configure squid to use no more than about 60 - 70% of the
available physical RAM-- ie, set the cache_mem parameter to about 2.5
or 3GB.


Better yet, don't run Squid at all.  It was designed for a computer
architecture that was already obsolete when Squid was first written.


This could be said of a lot of software, including many Unix  
flavors.  :-)


I can think of several things to criticise about Squid-- a config  
file which falls between Apache's httpd.conf and a sendmail.cf in  
terms of complexity is probably close to the top of my list, but for  
the simple purpose of saving limited network bandwidth using on-disk  
or in-memory caching, squid does just fine.  I'd be happy to look at  
Varnish when I get a chance, though.



It wouldn't be unreasonable to limit datasize to 3 GB on such a
machine, assuming that nothing you run will ever need to grow
larger...


...actually, maxdsiz is meaningless in FreeBSD 7, because the new
allocator uses mmap(2) instead of brk(2) / sbrk(2), so malloc() counts
towards the resident set size (ulimit -m), not the data segment size
(ulimit -d).


OK.  Nicole, the OP, mentioned amd64, not -CURRENT, but I'll keep  
this in mind for future reference.


Regards,
--
-Chuck

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-22 Thread Ganbold

Dag-Erling Smørgrav wrote:

Ganbold [EMAIL PROTECTED] writes:
  

Dag-Erling Smørgrav [EMAIL PROTECTED] writes:


Better yet, don't run Squid at all.
  

Ok, then what do you recommend instead of Squid?



That depends on what you use it for...

DES
  


We are using several squid machines (6 machines, each have all others as 
a siblings) for transparent caching/proxying

using gre tunnel and wccp2 (with Cisco router).
Can varnish work in such situation?

thanks,

Ganbold


--
A [golf] ball hitting a tree shall be deemed not to have hit the tree. 
Hitting a tree is simply bad luck and has no place in a scientific game. 
The player should estimate the distance the ball would have traveled if 
it had not hit the tree and play the ball from there, preferably atop a 
nice firm tuft of grass. -- Donald A. Metz

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-21 Thread N. Harrington

Hello
 I feel stupid, but I am confused about kern.maxdsiz (or datasize via limits 
command) on FreeBSD
amd64.

 I have seen many posts and suggestions to raise it to 1G. However it seems 
this only applies to
i386. By default, on servers I have with 4G of physical memory, and 2X that of 
swap, I am showing
a reported datasize limit of 33554432KB. far in excess of even my physical and 
swap combined!
 I have seen suggestions from mysql for maxdsize to be set to 1G. Obviously no 
such problem with
amd64?
 
 Is this just a high number chosen to let things run wild? (basically 
unlimited) I have been
having some problems with running squid and my servers locking up. I think, 
from the process
exceeding my physical memory and the server getting very unhappy trying use so 
much swap. (since
it seems the process size is so unlimited) 

 Could anyone help shed some light on this for me? If I have a server with 4G 
of memory, what
would be a safe /sane allowable maximum for datasize? (assuming a light 
networking load)

 Thanks!

  Nicole

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: What is a sane setting for maxdsize when running amd64? it seems many normal suggestions do not apply.

2007-08-21 Thread Chuck Swiger

On Aug 21, 2007, at 4:54 PM, N. Harrington wrote:
I have seen many posts and suggestions to raise it to 1G. However  
it seems this only applies to
i386. By default, on servers I have with 4G of physical memory, and  
2X that of swap, I am showing
a reported datasize limit of 33554432KB. far in excess of even my  
physical and swap combined!


True...that a big part of what you gain from running in 64-bit mode--  
a huge address space.


 I have seen suggestions from mysql for maxdsize to be set to 1G.  
Obviously no such problem with

amd64?


Right.

 Is this just a high number chosen to let things run wild?  
(basically unlimited) I have been
having some problems with running squid and my servers locking up.  
I think, from the process
exceeding my physical memory and the server getting very unhappy  
trying use so much swap. (since

it seems the process size is so unlimited)


You should configure squid to use no more than about 60 - 70% of the  
available physical RAM-- ie, set the cache_mem parameter to about 2.5  
or 3GB.


 Could anyone help shed some light on this for me? If I have a  
server with 4G of memory, what
would be a safe /sane allowable maximum for datasize? (assuming a  
light networking load)


It wouldn't be unreasonable to limit datasize to 3 GB on such a  
machine, assuming that nothing you run will ever need to grow larger...


--
-Chuck

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]