VMware Cloud Community
afrSupport
Contributor
Contributor

ql2xmaxqdepth=64 not working

After adding the above option to the /etc/vmware/esx.conf file and reboot the server, I still get the following in /var/log/vmkernel:

/var/log/vmkernel.1:Apr 14 12:36:11 esx2 vmkernel: 0:00:01:12.140 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:0:0 is 32

/var/log/vmkernel.1:Apr 14 12:36:11 esx2 vmkernel: 0:00:01:12.142 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:0:1 is 32

/var/log/vmkernel.1:Apr 14 12:36:11 esx2 vmkernel: 0:00:01:12.143 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:0:3 is 32

/var/log/vmkernel.1:Apr 14 12:36:11 esx2 vmkernel: 0:00:01:12.144 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:0:5 is 32

/var/log/vmkernel.1:Apr 14 12:36:11 esx2 vmkernel: 0:00:01:12.145 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:0:7 is 32

/var/log/vmkernel.1:Apr 14 12:36:11 esx2 vmkernel: 0:00:01:12.146 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:0:9 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.146 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:2 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.147 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:4 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.148 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:6 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.149 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:8 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.150 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:10 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.151 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:11 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.152 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:12 is 32

/var/log/vmkernel.1:Apr 14 12:36:12 esx2 vmkernel: 0:00:01:12.153 cpu14:1045)LinSCSI: 689: Queue depth for device vmhba0:2:13 is 32

Just wondering if there is something else I need to modify to make this \*stick*.

Reply
0 Kudos
15 Replies
admin
Immortal
Immortal

Show me the esx.conf section you changed

Reply
0 Kudos
afrSupport
Contributor
Contributor

Thanks for the quick reply... see below (there are two dual-port HBAs):

/device/001:00.0/class = "0c0400"

/device/001:00.0/devID = "2432"

/device/001:00.0/irq = "16"

/device/001:00.0/name = "QLogic Corp QLA2432 (rev 02)"

/device/001:00.0/options = "ql2xmaxqdepth=64"[/b]

/device/001:00.0/owner = "vmkernel"

/device/001:00.0/subsysDevID = "0138"

/device/001:00.0/subsysVendor = "1077"

/device/001:00.0/vendor = "1077"

/device/001:00.0/vmkname = "vmhba0"

/device/001:00.1/class = "0c0400"

/device/001:00.1/devID = "2432"

/device/001:00.1/irq = "17"

/device/001:00.1/name = "QLogic Corp QLA2432 (rev 02)"

/device/001:00.1/options = "ql2xmaxqdepth=64"[/b]

/device/001:00.1/owner = "vmkernel"

/device/001:00.1/subsysDevID = "0138"

/device/001:00.1/subsysVendor = "1077"

/device/001:00.1/vendor = "1077"

/device/001:00.1/vmkname = "vmhba1"

/device/004:00.0/class = "0c0400"

/device/004:00.0/devID = "2432"

/device/004:00.0/irq = "16"

/device/004:00.0/name = "QLogic Corp QLA2432 (rev 02)"

/device/004:00.0/options = "ql2xmaxqdepth=64"[/b]

/device/004:00.0/owner = "vmkernel"

/device/004:00.0/subsysDevID = "0138"

/device/004:00.0/subsysVendor = "1077"

/device/004:00.0/vendor = "1077"

/device/004:00.0/vmkname = "vmhba2"

/device/004:00.1/class = "0c0400"

/device/004:00.1/devID = "2432"

/device/004:00.1/irq = "17"

/device/004:00.1/name = "QLogic Corp QLA2432 (rev 02)"

/device/004:00.1/options = "ql2xmaxqdepth=64"[/b]

/device/004:00.1/owner = "vmkernel"

/device/004:00.1/subsysDevID = "0138"

/device/004:00.1/subsysVendor = "1077"

/device/004:00.1/vendor = "1077"

/device/004:00.1/vmkname = "vmhba3"

Reply
0 Kudos
admin
Immortal
Immortal

It looks like your syntax is correct - providing that 2nd character is an L not a one.

I did notice that these are QLA2432 and these are not on the HCL, I wonder if this has anything to do with it.

Reply
0 Kudos
admin
Immortal
Immortal

Disregard my 2nd comment perhaps it is related to This bug

Reply
0 Kudos
afrSupport
Contributor
Contributor

Indeed. Do you think I should open a support call?

Reply
0 Kudos
admin
Immortal
Immortal

This is behavior I have not seen before.

I am pretty sure it will not solve this - but I am clutching here.

I always change the Disk.SchedNumReqOutstanding in conjunction with setting queue depth. Did you also change this advanced parameter to 64?

This is in the advanced configuration options for the host under disk.

As far as an SR is concerned, this would be up to you and your timeframe. Perhaps on Monday when more people are back on the board, someone will have seen this.

Message was edited by:

BrianG

Reply
0 Kudos
afrSupport
Contributor
Contributor

The cards are indeed 2462 models.

\# cat /proc/scsi/qla2300/0

QLogic PCI to Fibre Channel Host Adapter for QLE2462:

Firmware version: 4.00.18, Driver version 7.07.04

Entry address = 0x7cd314

Request Queue = 0xce403000, Response Queue = 0xce414000

Request Queue count= 512, Response Queue count= 512

Total number of active commands = 0

Total number of interrupts = 7460

Total number of active IP commands = 0

Total number of IOCBs (used/max) = (0/0)

Total number of queued commands = 0

Device queue depth = 0x20

Number of free request entries = 100

Number of mailbox timeouts = 0

Number of ISP aborts = 0

Number of loop resyncs = 1

Number of retries for empty slots = 0

Number of reqs in pending_q= 0, retry_q= 0, done_q= 0, scsi_retry_q= 0

Host adapter:loop state= , flags= 0x70013

Dpc flags = 0x0

MBX flags = 0x0

SRB Free Count = 4096

Link down Timeout = 045

Port down retry = 015

Login retry count = 015

Loop down time = 120

Loop down abort time = 075

Commands retried with dropped frame(s) = 0

Configured data rate: 4 Gb/sec

Reply
0 Kudos
afrSupport
Contributor
Contributor

Additional info:

\# vmkload_mod -l | grep qla

qla2300_707 0x7cd000 0x44000 0x1f13450 0x72000 3 Yes

Seem that the qla2300_707[/b] module is being loaded instead of the qla2x00_707[/b] module that page 5 of the I/O Compatibility Guide For ESX Server 3.x indicates should be used.

Reply
0 Kudos
dheerajms
Enthusiast
Enthusiast

After changing esx.conf, did you execute 'esxcfg-boot -b' and check? New boot image has to have that changed information before it can accept your request.

Reply
0 Kudos
christianZ
Champion
Champion

Reply
0 Kudos
grasshopper
Virtuoso
Virtuoso

Since the esxcfg-boot command has a nasty reputation for killing your network config, I would advise running the vm-support script and optionally, taking screenshots of the host's network setup via VIC prior to running it.

Reply
0 Kudos
afrSupport
Contributor
Contributor

This was the ticket: http://www.vmware.com/community/thread.jspa?messageID=451432&#451432

Specifically the "[b]esxcfg-module -s ql2xmaxqdepth=64 qla2300_707[/b]" command.

Thanks to all for taking your time to assist.

Reply
0 Kudos
jbenard
Contributor
Contributor

I'M a newbie with VMware and VI3... I'm using QLA2432 ...

What is the exactly purpose to add this command esxcfg-module -s ql2xmaxqdepth=64 qla2300_707 ?

Thanx all.

Reply
0 Kudos
grasshopper
Virtuoso
Virtuoso

What is the exactly purpose to add this command

esxcfg-module -s ql2xmaxqdepth=64 qla2300_707 ?

The esxcfg-module command discussed in this thread is used to achieve results similar to the old school performance tweaks for ESX 2.x (KB's 1267 through 1269).

Queue Depth for ESX 2.x:[/b]

KB1267[/url]

KB1268[/url]

KB1269[/url]

Queue Depth for ESX 3.x:[/b]

For the new info on doing this with ESX 3.x, see pages 107-109:

SAN_Configuration_Guide[/url].

Reply
0 Kudos
CyberV
Contributor
Contributor

When cloning a VM or just create a new Template on my ESX 3.0.1 attached to a MSA1000 take very long time. More then 2 or 3 hours sometimes. It's normal ? It's something that I can change to fixed this issue ?

Thanx for your help.

Reply
0 Kudos