On Fri, Apr 25, 2014 at 10:52 AM, Gerrit Kühn wrote:
>
>
> The irq-igb1 clibs to top every now and then (always less than 1% cpu
> usage though), but there is never much cpu usage shown for any task.
> However, the overall load is significant. What might the system be doing?
>
>
http://lists.freebs
- Original Message -
From: "Gerrit Kühn"
To: "Steven Hartland"
Cc:
Sent: Friday, April 25, 2014 4:52 PM
Subject: Re: NFS over LAGG / lacp poor performance
On Fri, 25 Apr 2014 15:55:29 +0100 "Steven Hartland"
wrote about Re: NFS over LAGG / lacp
On Fri, 25 Apr 2014 15:55:29 +0100 "Steven Hartland"
wrote about Re: NFS over LAGG / lacp poor
performance:
SH> Run top then enable system processes (shift+S) and threads (shift+H)
SH> then disable idle process display (z) if your seeing significant CPU
SH> usage to the igb
- Original Message -
From: "Gerrit Kühn"
On Fri, 25 Apr 2014 15:25:04 +0100 "Steven Hartland"
wrote about Re: NFS over LAGG / lacp poor
performance:
SH> We saw the issue with 10-RELEASE last weekend on a machine with 6 x
SH> igb's where by th
On Fri, 25 Apr 2014 15:25:04 +0100 "Steven Hartland"
wrote about Re: NFS over LAGG / lacp poor
performance:
SH> We saw the issue with 10-RELEASE last weekend on a machine with 6 x
SH> igb's where by the system was burning CPU in the interrupt handlers,
SH> setting num_que
- Original Message -
From: "Gerrit Kühn"
To: "Steven Hartland"
Cc:
Sent: Friday, April 25, 2014 3:13 PM
Subject: Re: NFS over LAGG / lacp poor performance
On Fri, 25 Apr 2014 15:02:15 +0100 "Steven Hartland"
wrote about Re: NFS over LAGG / lacp
On Fri, 25 Apr 2014 15:02:15 +0100 "Steven Hartland"
wrote about Re: NFS over LAGG / lacp poor
performance:
SH> We find that large numbers of queues causes high interrupt issues
Like the thing I am seeing with igb1 on my system?
---
root@storage:/root #vmstat
W dniu 2014-04-25 16:02, Steven Hartland pisze:
> We find that large numbers of queues causes high interrupt issues
> however at a guess you did this to enable the machine to boot with
> all nics due to lack of auto mbuf tuning in 9.x.
>
> I'd go with ~2 queues per nic.
Yes, probably you're right
- Original Message -
From: "Marek Salwerowicz"
To: "Steven Hartland" ; "Gerrit Kühn"
Cc:
Sent: Friday, April 25, 2014 2:57 PM
Subject: Re: NFS over LAGG / lacp poor performance
W dniu 2014-04-25 15:27, Steven Hartland pisze:
In that case I believ
W dniu 2014-04-25 15:27, Steven Hartland pisze:
> In that case I believe you've hard coded the number of queues, check
> /boot/loader.conf
> for references to this.
Yes, that's true:
% cat /boot/loader.conf
debug.acpi.max_tasks="128"
if_lagg_load="YES"
kern.ipc.nmbclusters="131072"
hw.
Steven Hartland wrote:
>
> - Original Message -
> From: "Marek Salwerowicz"
> To: "Steven Hartland" ; "Gerrit Kühn"
>
> Cc:
> Sent: Friday, April 25, 2014 2:06 PM
> Subject: Re: NFS over LAGG / lacp poor performance
&
- Original Message -
From: "Marek Salwerowicz"
To: "Steven Hartland" ; "Gerrit Kühn"
Cc:
Sent: Friday, April 25, 2014 2:06 PM
Subject: Re: NFS over LAGG / lacp poor performance
W dniu 2014-04-25 14:55, Steven Hartland pisze:
- Original Message
Marek Salwerowicz wrote:
> W dniu 2014-04-25 13:48, Rick Macklem pisze:
> > Well, you don't mention what command(s) you are using to transfer
> > the
> > data, but I would guess you have one serial data transfer for each
> > command.
> > (Put another way, if you are only running one command to tran
W dniu 2014-04-25 14:55, Steven Hartland pisze:
> - Original Message - From: "Marek Salwerowicz"
>
>
>> W dniu 2014-04-25 14:01, Gerrit Kühn pisze:
>>> Thanks for your input. As far as I understood so far, there should
>>> be one
>>> igb queue created per cpu core in the system by default
- Original Message -
From: "Marek Salwerowicz"
W dniu 2014-04-25 14:01, Gerrit Kühn pisze:
Thanks for your input. As far as I understood so far, there should be one
igb queue created per cpu core in the system by default (and this is what
I see on my system). But my irq rate looks qu
W dniu 2014-04-25 13:57, Marek Salwerowicz pisze:
> (executed on storage1)
> # cp -a file1 /net/storage2/ &
> # cp -a file2 /net/storage2/ &
> # cp -a file3 /net/storage2/ &
> # cp -a file4 /net/storage2/ &
Just to add:
both source and destination are based on ZFS raidz2 pool with mirrored
lo
W dniu 2014-04-25 14:01, Gerrit Kühn pisze:
> Thanks for your input. As far as I understood so far, there should be one
> igb queue created per cpu core in the system by default (and this is what
> I see on my system). But my irq rate looks quite high to me (and it is
> only on one of these queues)
On Fri, 25 Apr 2014 13:34:06 +0200 Marek Salwerowicz
wrote about Re: NFS over LAGG / lacp poor performance:
GK> irq256: igb0:que 0 99396134 64
GK> irq257: igb0:que 1 61496018 39
GK> irq258: igb0:que 2 101687742 66
GK>
W dniu 2014-04-25 13:48, Rick Macklem pisze:
> Well, you don't mention what command(s) you are using to transfer the
> data, but I would guess you have one serial data transfer for each command.
> (Put another way, if you are only running one command to transfer the data,
> there will only be one
Marek Salwerowicz wrote:
> Hi list,
>
> I have two FreeBSD boxes (both based on SuperMicro X9DRD-7LN4F-JBOD
> motherboard, with 32GB RAM, 1 CPU :Intel(R) Xeon(R) CPU E5-2640 v2)
>
> storage1% uname -a
> FreeBSD storage1 9.1-RELEASE-p10 FreeBSD 9.1-RELEASE-p10 #0: Sun Jan
> 12
> 20:11:23 UTC 2014
- Original Message -
From: "Robert Blacquiere"
Hi Marek,
On Fri, Apr 25, 2014 at 09:48:36AM +0200, Marek Salwerowicz wrote:
Hi list,
Both boxes are connected to the same switch (HP 1910-48G)
I need to transfer around 10 TB of data from storage1 to storage2
I obeserve that during
Hello Gerrit,
W dniu 2014-04-25 11:37, Gerrit Kühn pisze:
> On Fri, 25 Apr 2014 09:48:36 +0200 Marek Salwerowicz
> wrote about NFS over LAGG / lacp poor performance:
>
>
>
> Could you have a look on your system and let me know how your interrupts
> are spread?
For me on
Hi Marek,
On Fri, Apr 25, 2014 at 09:48:36AM +0200, Marek Salwerowicz wrote:
> Hi list,
>
> Both boxes are connected to the same switch (HP 1910-48G)
>
> I need to transfer around 10 TB of data from storage1 to storage2
> I obeserve that during copying, only one NIC (instead of all 4) is used
On Fri, 25 Apr 2014 09:48:36 +0200 Marek Salwerowicz
wrote about NFS over LAGG / lacp poor performance:
Hello Marek,
MS> FreeBSD storage1 9.1-RELEASE-p10 FreeBSD 9.1-RELEASE-p10 #0: Sun Jan 12
MS> 20:11:23 UTC 2014
MS> r...@amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENER
Hi list,
I have two FreeBSD boxes (both based on SuperMicro X9DRD-7LN4F-JBOD
motherboard, with 32GB RAM, 1 CPU :Intel(R) Xeon(R) CPU E5-2640 v2)
storage1% uname -a
FreeBSD storage1 9.1-RELEASE-p10 FreeBSD 9.1-RELEASE-p10 #0: Sun Jan 12
20:11:23 UTC 2014
25 matches
Mail list logo
