* RE: I/O Blocking
2004-07-22 16:29 I/O Blocking Jason C. Leach
@ 2004-07-22 18:02 ` Guy
2004-07-22 18:09 ` Jason C. Leach
2004-07-22 19:51 ` Kanoa Withington
2004-07-23 1:47 ` berk walker
2 siblings, 1 reply; 6+ messages in thread
From: Guy @ 2004-07-22 18:02 UTC (permalink / raw)
To: 'Jason C. Leach', linux-raid
What is the CPU load like? Are you at 100%? If so a second CPU may help.
If you are blocked on i/o it should not use much CPU, so my guess is your
CPU load is very low, and a second CPU won't do much. How much RAM do you
have? Are you swapping?
Guy
-----Original Message-----
From: linux-raid-owner@vger.kernel.org
[mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Jason C. Leach
Sent: Thursday, July 22, 2004 12:29 PM
To: linux-raid@vger.kernel.org
Subject: I/O Blocking
Hi,
Both the RAID (Promise SuperTrak 6000) or the USB subsystems really
drive the system load up. I'm using a AMD 3200+ system that is only a
few months old with the 2.4.25 kernel. When the I/O systems are
blocking the system is very unresponsive. Some times samba will timeout,
or I'll wait several seconds for commands like df, top, w and so on.
I am wondering if I were to use a dual CPU system, would this solve my
I/O blocking problem? Would one CPU tend to the I/O subsystem and the
second would deal with other system requests I (samba, DB, LDAP, ...)?
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: I/O Blocking
2004-07-22 18:02 ` Guy
@ 2004-07-22 18:09 ` Jason C. Leach
2004-07-22 18:37 ` Guy
0 siblings, 1 reply; 6+ messages in thread
From: Jason C. Leach @ 2004-07-22 18:09 UTC (permalink / raw)
To: Guy; +Cc: linux-raid
Guy:
CPU load is very low (95-96% idle), load avg is 2. The system has 512MB
RAM, some free and only a few megs swapped out. If I was really working
the RAID card I would get load avgs more like 2.5-3. Right now just
light USB and system usage and I am at 2.
I though if one CPU was tied up with the process that is blocking (using
the RAID card), then the second would be free to deal with user
requests. Or am I misunderstanding SMP and blocking I/O?
J.
Guy wrote:
>What is the CPU load like? Are you at 100%? If so a second CPU may help.
>If you are blocked on i/o it should not use much CPU, so my guess is your
>CPU load is very low, and a second CPU won't do much. How much RAM do you
>have? Are you swapping?
>
>Guy
>
>-----Original Message-----
>From: linux-raid-owner@vger.kernel.org
>[mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Jason C. Leach
>Sent: Thursday, July 22, 2004 12:29 PM
>To: linux-raid@vger.kernel.org
>Subject: I/O Blocking
>
>Hi,
>
>Both the RAID (Promise SuperTrak 6000) or the USB subsystems really
>drive the system load up. I'm using a AMD 3200+ system that is only a
>few months old with the 2.4.25 kernel. When the I/O systems are
>blocking the system is very unresponsive. Some times samba will timeout,
>or I'll wait several seconds for commands like df, top, w and so on.
>
>I am wondering if I were to use a dual CPU system, would this solve my
>I/O blocking problem? Would one CPU tend to the I/O subsystem and the
>second would deal with other system requests I (samba, DB, LDAP, ...)?
>-
>To unsubscribe from this list: send the line "unsubscribe linux-raid" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>
>
^ permalink raw reply [flat|nested] 6+ messages in thread
* RE: I/O Blocking
2004-07-22 18:09 ` Jason C. Leach
@ 2004-07-22 18:37 ` Guy
0 siblings, 0 replies; 6+ messages in thread
From: Guy @ 2004-07-22 18:37 UTC (permalink / raw)
To: 'Jason C. Leach'; +Cc: linux-raid
If your 1 CPU is 95% idle, it can do other "user requests". But not if the
other "user requests" require disk I/O, they will also block regardless of
which CPU they are using. I think there is a performance problem with your
disks. But just a guess.
Guy
-----Original Message-----
From: linux-raid-owner@vger.kernel.org
[mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Jason C. Leach
Sent: Thursday, July 22, 2004 2:10 PM
To: Guy
Cc: linux-raid@vger.kernel.org
Subject: Re: I/O Blocking
Guy:
CPU load is very low (95-96% idle), load avg is 2. The system has 512MB
RAM, some free and only a few megs swapped out. If I was really working
the RAID card I would get load avgs more like 2.5-3. Right now just
light USB and system usage and I am at 2.
I though if one CPU was tied up with the process that is blocking (using
the RAID card), then the second would be free to deal with user
requests. Or am I misunderstanding SMP and blocking I/O?
J.
Guy wrote:
>What is the CPU load like? Are you at 100%? If so a second CPU may help.
>If you are blocked on i/o it should not use much CPU, so my guess is your
>CPU load is very low, and a second CPU won't do much. How much RAM do you
>have? Are you swapping?
>
>Guy
>
>-----Original Message-----
>From: linux-raid-owner@vger.kernel.org
>[mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Jason C. Leach
>Sent: Thursday, July 22, 2004 12:29 PM
>To: linux-raid@vger.kernel.org
>Subject: I/O Blocking
>
>Hi,
>
>Both the RAID (Promise SuperTrak 6000) or the USB subsystems really
>drive the system load up. I'm using a AMD 3200+ system that is only a
>few months old with the 2.4.25 kernel. When the I/O systems are
>blocking the system is very unresponsive. Some times samba will timeout,
>or I'll wait several seconds for commands like df, top, w and so on.
>
>I am wondering if I were to use a dual CPU system, would this solve my
>I/O blocking problem? Would one CPU tend to the I/O subsystem and the
>second would deal with other system requests I (samba, DB, LDAP, ...)?
>-
>To unsubscribe from this list: send the line "unsubscribe linux-raid" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: I/O Blocking
2004-07-22 16:29 I/O Blocking Jason C. Leach
2004-07-22 18:02 ` Guy
@ 2004-07-22 19:51 ` Kanoa Withington
2004-07-23 1:47 ` berk walker
2 siblings, 0 replies; 6+ messages in thread
From: Kanoa Withington @ 2004-07-22 19:51 UTC (permalink / raw)
To: Jason C. Leach; +Cc: linux-raid
Jason,
There is a technical answer to your question which I cannot properly
relate to you. The practical answer is yes, an SMP system will
generally perform better under heavy interrupt conditions and would
probably help with the situation you are having.
That said, the problem you are having is not lack of CPU
performance. I don't know what it is but you need to look into it
before throwing another mega-cpu at it. You might have too many I/O
devices sharing the same IRQ, it could be something weird like a
haywire parallel port device. Take a careful look at how your
equipment is set up and do some experiments turning things off and on,
etc. You can probably get to the bottom of it without upgrading
anything.
-Kanoa
On Thu, 22 Jul 2004, Jason C. Leach wrote:
> Hi,
>
> Both the RAID (Promise SuperTrak 6000) or the USB subsystems really
> drive the system load up. I'm using a AMD 3200+ system that is only a
> few months old with the 2.4.25 kernel. When the I/O systems are
> blocking the system is very unresponsive. Some times samba will timeout,
> or I'll wait several seconds for commands like df, top, w and so on.
>
> I am wondering if I were to use a dual CPU system, would this solve my
> I/O blocking problem? Would one CPU tend to the I/O subsystem and the
> second would deal with other system requests I (samba, DB, LDAP, ...)?
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: I/O Blocking
2004-07-22 16:29 I/O Blocking Jason C. Leach
2004-07-22 18:02 ` Guy
2004-07-22 19:51 ` Kanoa Withington
@ 2004-07-23 1:47 ` berk walker
2 siblings, 0 replies; 6+ messages in thread
From: berk walker @ 2004-07-23 1:47 UTC (permalink / raw)
To: Jason C. Leach; +Cc: linux-raid
Jason C. Leach wrote:
> Hi,
>
> Both the RAID (Promise SuperTrak 6000) or the USB subsystems really
> drive the system load up. I'm using a AMD 3200+ system that is only a
> few months old with the 2.4.25 kernel. When the I/O systems are
> blocking the system is very unresponsive. Some times samba will
> timeout, or I'll wait several seconds for commands like df, top, w and
> so on.
>
> I am wondering if I were to use a dual CPU system, would this solve my
> I/O blocking problem? Would one CPU tend to the I/O subsystem and the
> second would deal with other system requests I (samba, DB, LDAP, ...)?
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
HMMM I seemed to have lost the 1st of the thread, 2 cents worth .. what
ya pay for ...
I might be supect of the m/b IDE channels (personally had lots of
failures) - SMART drives that are not configured as SMART (sorry, I
always use Intel, AMD...?) enableing in bios and kernel or mod might
give clues. I, also, think that there must be an IRQ conflict, IRQ
unanswered, or same ilk.
You could test by swapping drives, swapping m/b's. What sizes are your
disks? What RAM do you use,, I do have a plethora of hwde I could loan.
b-
^ permalink raw reply [flat|nested] 6+ messages in thread