Hi Dan,

i just talked about openmosix because i read about
shared memory segments. my only desire is to make mysql
able to spawn 1000-1500 connections and *not* to tell
my client "cant create new thread, perhaps out of memory"
while 1.5 GB of RAM is only used for caching. Is that really
possible that mysql is not able to handle this amount of connections ?

My best Regards,
Philipp


On Wed, 19 Mar 2003 10:01:38 -0600
Dan Nelson <[EMAIL PROTECTED]> wrote:

> In the last episode (Mar 19), Philipp said:
> > thank you for your answer. While i was reading about clustering some
> > weeks ago i read the openmosix FAQ claiming that openmosix would not
> > work with apache, because apache was using shared memory to
> > communicate with its threads, and i always thought mysql is designed
> > the same way. Please someone correct me if i am wrong.
> 
> SYSV shared memory (shm*) is a block of memory that one process
> creates, and depending on the access flags, multiple processes can
> attach to and see each other's changes.
> 
> Threads use shared memory by definition, because a threaded application
> is still one process.  No SYSV shm tricks are needed.
> 
> Openmosix won't be able to balance mysql threads, because even if it
> were possible to synchronize shared memory between machines with Mosix,
> synching thread mutexes would be horrendously slow.  Better to just get
> a multi-CPU box.
> 
> -- 
>       Dan Nelson
>       [EMAIL PROTECTED]
> 


-- 


-----------------
Philipp Steinkrueger
Oberberg Online Informationssysteme GmbH
Technik
http://www.oberberg.net

PGPkeyID: 690A9504
Key Fingerprint: 35CE 467E C813 06B0 B8E3  0275 2B1E E84A 690A 9504

---------------------------------------------------------------------
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/           (the list archive)

To request this thread, e-mail <[EMAIL PROTECTED]>
To unsubscribe, e-mail <[EMAIL PROTECTED]>
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

Reply via email to