[Mailman-Users] Slow command-line unsubscription

[Charlie Watts]

I'm trying to unsubscribe about 1000 addresses from a list of about
12,000.

This is on FreeBSD 4.3, a dual-proc PIII 500. Otherwise un-loaded.

Command-line:
~/bin/remove_members -f /tmp/un-newsletter newsletter

It is taking  -forever- . Watching ~logs/subscribe, addresses are being
un-subscribed at about 1 message per second.

Is this normal?

-- 
Charlie Watts
[EMAIL PROTECTED]Frontier Internet
Systems Janitor and Network Plumber   http://www.frontier.net/


--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users

Re: [Mailman-Users] Slow command-line unsubscription

[Chuq Von Rospach]

On 7/13/01 3:43 PM, "Charlie Watts" <[EMAIL PROTECTED]> wrote:

> I'm trying to unsubscribe about 1000 addresses from a list of about
> 12,000.

> It is taking  -forever- . Watching ~logs/subscribe, addresses are being
> un-subscribed at about 1 message per second.
> 
> Is this normal?

Wouldn't surprise me. Since it's a single process, your MP machine isn't
going to use all it's CPU, but the operation is going to be primarily
disk-limted anyway -- are you using fast disks? Slow disks?


-- 
Chuq Von Rospach, Internet Gnome 
[<[EMAIL PROTECTED]> = <[EMAIL PROTECTED]> = <[EMAIL PROTECTED]>]
Yes, yes, I've finally finished my home page. Lucky you.

The first rule of holes: If you are in one, stop digging.




--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users

Re: [Mailman-Users] Slow command-line unsubscription

[J C Lawrence]

On Fri, 13 Jul 2001 22:56:13 -0700 
Chuq Von Rospach <[EMAIL PROTECTED]> wrote:

> On 7/13/01 3:43 PM, "Charlie Watts" <[EMAIL PROTECTED]> wrote:

>> I'm trying to unsubscribe about 1000 addresses from a list of
>> about 12,000.

>> It is taking -forever- . Watching ~logs/subscribe, addresses are
>> being un-subscribed at about 1 message per second.

>> Is this normal?

> Wouldn't surprise me. Since it's a single process, your MP machine
> isn't going to use all it's CPU, but the operation is going to be
> primarily disk-limted anyway -- are you using fast disks? Slow
> disks?

According to brief examination here its lock bound rather than IO or
CPU.

-- 
J C Lawrence   ("`-''-/").___..--''"`-._ 
-(*) `6_ 6  )   `-.  ( ).`-.__.`)
[EMAIL PROTECTED]   (_Y_.)'  ._   )  `._ `. ``-..-'  
http://www.kanga.nu/~claw/_..`--'_..-_/  /--'_.' ,' 
I never claimed to be human (il),-''  (li),'  ((!.-'   

--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users

Re: [Mailman-Users] Slow command-line unsubscription

[Charlie Watts]

On Sat, 14 Jul 2001, J C Lawrence wrote:
> On Fri, 13 Jul 2001 22:56:13 -0700
> Chuq Von Rospach <[EMAIL PROTECTED]> wrote:
>
> > On 7/13/01 3:43 PM, "Charlie Watts" <[EMAIL PROTECTED]> wrote:
>
> >> I'm trying to unsubscribe about 1000 addresses from a list of
> >> about 12,000.
>
> >> It is taking -forever- . Watching ~logs/subscribe, addresses are
> >> being un-subscribed at about 1 message per second.
>
> >> Is this normal?
>
> > Wouldn't surprise me. Since it's a single process, your MP machine
> > isn't going to use all it's CPU, but the operation is going to be
> > primarily disk-limted anyway -- are you using fast disks? Slow
> > disks?
>
> According to brief examination here its lock bound rather than IO or
> CPU.

Thank you both. Chuq - yes, fairly fast disks, but even really slow disks
should be able to change a (small) file faster than that.

The lock-bound explanation makes complete sense.

It wasn't so much a -problem- that it took a long time, as that it seemed
potentially broken, and I wanted to check.

-- 
Charlie Watts
[EMAIL PROTECTED]Frontier Internet
Systems Janitor and Network Plumber   http://www.frontier.net/


--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users

Re: [Mailman-Users] Slow command-line unsubscription

[Chuq Von Rospach]

On 7/14/01 11:32 AM, "J C Lawrence" <[EMAIL PROTECTED]> wrote:

> According to brief examination here its lock bound rather than IO or
> CPU.

So mailman is setting and removing the lock for every address? If so --
Barry, isn't that a design flaw for this case?

-- 
Chuq Von Rospach, Internet Gnome 
[<[EMAIL PROTECTED]> = <[EMAIL PROTECTED]> = <[EMAIL PROTECTED]>]
Yes, yes, I've finally finished my home page. Lucky you.

"He doesn't have ulcers, but he's a carrier."




--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users

Re: [Mailman-Users] Slow command-line unsubscription

[Harold Paulson]

Hi,

more than being disk or CPU bound, isn't this mostly, 
have-to-send-an-email-to-each-address bound?

- H


At 4:43 PM -0600 7/13/01, Charlie Watts wrote:
>I'm trying to unsubscribe about 1000 addresses from a list of about
>12,000.
>
>This is on FreeBSD 4.3, a dual-proc PIII 500. Otherwise un-loaded.
>
>Command-line:
>~/bin/remove_members -f /tmp/un-newsletter newsletter
>
>It is taking  -forever- . Watching ~logs/subscribe, addresses are being
>un-subscribed at about 1 message per second.
>
>Is this normal?
>
>--
>Charlie Watts
>[EMAIL PROTECTED]Frontier Internet
>Systems Janitor and Network Plumber   http://www.frontier.net/

-- 

Harold Paulson  Sierra Web Design
[EMAIL PROTECTED]   http://www.sierraweb.com
VOICE: 775.833.9500 FAX:   810.314.1517

--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users

RE: [Mailman-Users] Slow command-line unsubscription

[Enriko Groen]

> -Original Message-
> From: Harold Paulson [mailto:[EMAIL PROTECTED]]

> more than being disk or CPU bound, isn't this mostly, 
> have-to-send-an-email-to-each-address bound?

No...

I did the same thing last thursday without email notification.
Unsubscribed 4000+ adresses which took about 20 minutes.

--
Enriko Groen, Hosting manager

netivity bv   www.netivity.nl   [EMAIL PROTECTED]
038 - 850 1000   van nagellstraat 4  8011 eb  zwolle

  

--
Mailman-Users maillist  -  [EMAIL PROTECTED]
http://mail.python.org/mailman/listinfo/mailman-users