[Linux-PowerEdge] R320 with PERC H710 and SSD?

[Gregory Gulik]

Is there a reason this configuration would cause problems?

R320 server running Windows Server 2012
PERC H710 Mini
Two Samsung EVO 850 in RAID-1

What we're seeing is very poor I/O performance where even copying files 
can be slow much of the time, but not always.  The I/O performance is 
extremely inconsistent.



--
Gregory Gulik

email | delivered 
<https://www.facebook.com/ReachMail><https://twitter.com/Reachmail><http://www.linkedin.com/company/reachmail-inc-> 


__
REACHMAIL Inc.  |  Main 888-947-3224  |  180 W. Adams St.  |  Suite 800 
|  Chicago, IL 60603
ggu...@reachmail.com <mailto:ggu...@reachmail.com> | Direct 312-229-0132 
| http://www.reachmail.com <http://www.reachmail.com/>


___
Linux-PowerEdge mailing list
Linux-PowerEdge@dell.com
https://lists.us.dell.com/mailman/listinfo/linux-poweredge

Re: Returning Network stability problems on R710 servers and BCM5709

[Gregory Gulik]

I too had this problem a few months ago when we put our first R410 into 
production.
The resolution was to upgrade the ethernet driver.  I got it from Dell's 
website then installed it on the CentOS system:


gtar xvzf Bcom_LAN_14.1.0_LinuxR5S10_DKMS_A01.tar.gz

Find the appropriate version:
Bcom_LAN_14.1.0_LinuxR5S10_DKMS_A01/NetXtremeII

Had to install some dependencies:
yum -y install rpm-build dkms

Then install the driver:
rpm -ihv netxtreme2-5.0.17-1.dkms.noarch.rpm


On 2/9/2010 3:25 PM, Big Wave Dave wrote:
> We had to do the same thing on R410's.  We ran into a situation where
> the one of the interfaces would stop working, and ethtool would show
> "no link".  Restarting network wouldn't fix anything, only a reboot
> would resolve the problem.  Moving to the Broadcomm provided driver
> has solved this for us.
>
> Dave
>
> On Wed, Feb 3, 2010 at 6:30 AM, Carlson, Timothy S
>   wrote:
>
>> I've moved away from the RHEL/Centos driver and have gone directly to the 
>> bnx2 driver from Broadcomm.
>>
>> dmesg | grep bnx
>> Broadcom NetXtreme II Gigabit Ethernet Driver bnx2 v1.9.20b (July 9, 2009)
>> bnx2: eth0: using MSI
>>
>> That driver seems stable for me. I was seeing your things similar to your 
>> problem and this driver fixed things right up for me.
>>
>> http://www.broadcom.com/support/ethernet_nic/driver-sla.php?driver=NX2-Linux
>>
>> You'll need to download that driver and rebuild it from the SRPM. You'll 
>> also need to rebuild the driver for each kernel update which is a pain.
>>
>> Tim
>>
>> -Original Message-
>> From: linux-poweredge-boun...@dell.com 
>> [mailto:linux-poweredge-boun...@dell.com] On Behalf Of James Sparenberg
>> Sent: Wednesday, February 03, 2010 1:08 AM
>> To: linux-poweredge@dell.com
>> Subject: Returning Network stability problems on R710 servers and BCM5709
>>
>> All,
>>
>>I'm referencing an earlier thread from last Sept.
>>
>> http://lists.us.dell.com/pipermail/linux-poweredge/2009-September/040252.html
>>
>>In it there was a discussion related to stability problems with the 
>> Broadcom BCM5709 on a Dell r610, where there would be a loss of connectivity 
>> for new connections but existing connections, or all connections of a 
>> different protocol passed.
>>
>> For example.  Just now I lost the ability to ping eth0, or get NIS 
>> authentication on that IP, I also lost the ability to get TFTP connections 
>> via the eth1 address.  However at the same time DHCP is running against 
>> eth1, and SNMP NTP and HTTP over port 1 (webmin) where merrily working 
>> quite well on eth0.
>>
>> OS CentOS 5.4
>>
>> kernel   2.6.18-164.10.1.el5 SMP x86_64
>> Kernel module bk2
>>
>> modinfo output
>>
>> filename:   
>> /lib/modules/2.6.18-164.10.1.el5.centos.plus/kernel/drivers/net/bnx2.ko
>> version:1.9.3
>> license:GPL
>> description:Broadcom NetXtreme II BCM5706/5708/5709/5716 Driver
>> author: Michael Chan
>> srcversion: 1040A42F87B8BE8A019736C
>> alias:  pci:v14E4d163Csv*sd*bc*sc*i*
>> alias:  pci:v14E4d163Bsv*sd*bc*sc*i*
>> alias:  pci:v14E4d163Asv*sd*bc*sc*i*
>> alias:  pci:v14E4d1639sv*sd*bc*sc*i*
>> alias:  pci:v14E4d16ACsv*sd*bc*sc*i*
>> alias:  pci:v14E4d16AAsv*sd*bc*sc*i*
>> alias:  pci:v14E4d16AAsv103Csd3102bc*sc*i*
>> alias:  pci:v14E4d164Csv*sd*bc*sc*i*
>> alias:  pci:v14E4d164Asv*sd*bc*sc*i*
>> alias:  pci:v14E4d164Asv103Csd3106bc*sc*i*
>> alias:  pci:v14E4d164Asv103Csd3101bc*sc*i*
>> depends:
>> vermagic:   2.6.18-164.10.1.el5.centos.plus SMP mod_unload gcc-4.1
>> parm:   disable_msi:Disable Message Signaled Interrupt (MSI) (int)
>> parm:   enable_entropy:Allow bnx2 to populate the /dev/random 
>> entropy pool (int)
>> module_sig: 
>> 883f3504b47af9bd3b84a368dd51f2112b6b90a0ed1bac15e1b94720602336594dc65775db83c460991575cc8694cf9c03aca6e623e0950281e5094
>>
>> So you can see that the version I have exceeds the version said to be stable 
>> in the prior thread.  BTW this chassis is about 1 month old so it should 
>> (but unverified) have the latest BIOS.
>>
>> Ironic part.  Same model running the same version/kernel of CentOS (kick 
>> start install so all my boxes are the same) is running some load testing 
>> pushing millions of sessions and billions (soaking 4 1G nics) of packets 
>> without a hitch in our LAB, testing out equipment, yet, this box which has a 
>> relatively low throughput is the one that locks up.
>>
>> Any thoughts or suggestions would be appreciated.  So far nothing in normal 
>> logs so I'm going to turn some additional logging on.
>>
>> James Sparenberg
>>
>> ___
>> Linux-PowerEdge mailing list
>> Linux-PowerEdge@dell.com
>> https://lists.us.dell.com/mailman/listinfo/linux-poweredge
>> Please read the FAQ at http://lists.u

E-mail server with two interfaces stops transmitting on one interface

[Gregory Gulik]

We have a mail server that sends out E-mail via two ISPs and each ISP is 
connected to the mail server through a different interface.  eth0 is on 
ISP1 and eth1 is on ISP2.  Each interface does go through a firewall so 
they are not directly on the Internet.  Server is running CentOS 5.3


The set-up of each is
eth0: 10.0.0.0/8
eth1: 172.16.0.0/12

There are dozens of IPs on each interface that are used as source IPs 
for the E-mail server.  The router on each ISP does the NAT translation 
to translate the internal IP address to the actual public IPs.


The routing table is set up as follows:

172.16.0.0/12 dev eth1  proto kernel  scope link  src 172.16.0.45
10.0.0.0/8 dev eth0  proto kernel  scope link  src 10.0.0.45
default via 10.0.0.1 dev eth0

To force E-mails with a source IP on the eth1 we have the following rule 
set up:


ip route add default via 172.16.0.1 dev eth1 table wan2
ip rule add from 172.16.0.0/12 table wan2


That seems to work fine for a while, E-mail goes out both interfaces 
with the correct source IPs and everyone seems happy.  However at some 
point for some reason outbound connections on eth1 time out and nothing 
we've tried gets them going again until a reboot after which everything 
works just fine again.  The entire time eth0 traffic is completely 
unaffected.


We verify nothing goes out with a command such as:
ncat -s 172.16.1.1 www.google.com  80

That command times out on all source IPs we've tried.

What are we missing?

--
Greg Gulik http://www.gulik.com/greg/
greg @ gulik.com


___
Linux-PowerEdge mailing list
Linux-PowerEdge@dell.com
https://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq