poor MSA performance

[Marius Pana_1]

Hi all,

I was at a clients site today where there is an MSA 1500CS and an HPdl380g5 running CentOS4. The client was complaining of poor performance and after doing a couple of tests we got the below results.

An hdparm ran against the msa looks like this:

[root@yaeyo var]# hdparm -t /dev/sda

/dev/sda:
Timing buffered disk reads: 116 MB in 3.07 seconds = 37.84 MB/sec


I have also noticed that a local drive that used to be somewhere over 90 mb
/sec (a 15k rpm scsi drive), after inserting the HBA in the server has
slowed down to this:

[root@yaeyo var]# hdparm -t /dev/cciss/c0d0

/dev/cciss/c0d0:
Timing buffered disk reads: 78 MB in 3.04 seconds = 25.68 MB/sec
HDIO_DRIVE_CMD(null) (wait for flush complete) failed: Inappropriate ioctl
for device


Of course we are talking about running the hdparm test agains _idle_ drives.
Other thing I have noticed is that withouth the HBA, hpacucli returns almost
immediate output from command "ctrl all show config" while after inserting
the HBA it takes about 6-7 seconds of screen freeze before it dumps the
requested command output.

Questions:
1) What is the expected hdparm -t result ran against the MSA1500CS array in
the idle mode and in the default configuration created by the cpqacuxe web
gui (A RAID 6 array of 6 SCSI ULTRA320 disks) ?
2) What do the following dmesg messages mean?

lpfc 0000:0a:01.0: 0:0326 Rsp Ring 0 error: IOCB Data: xff000020 x22580c0 x0
x0 xe3 x0 x3d60ce1 xa29b16
lpfc 0000:0a:01.0: 0:0729 FCP cmd xa3 failed <0/1> status: x1 result: xe3
Data: x3d6 xce1
lpfc 0000:0a:01.0: 0:0730 FCP command failed: RSP Data: x9 x0 xe3 x0 x8 x0
lpfc 0000:0a:01.0: 0:0716 FCP Read Underrun, expected 255, residual 227
Data: xff xa3 x0
lpfc 0000:0a:01.0: 0:0326 Rsp Ring 0 error: IOCB Data: xff000020 x22580c0 x0
x0 xab x0 x3d40ce2 xa29b16
lpfc 0000:0a:01.1: 1:0326 Rsp Ring 0 error: IOCB Data: xff000020 x23301c8 x0
x0 xe3 x0 x1e8083a xa29b16
lpfc 0000:0a:01.1: 1:0729 FCP cmd xa3 failed <0/3> status: x1 result: xe3
Data: x1e8 x83a
lpfc 0000:0a:01.1: 1:0730 FCP command failed: RSP Data: x9 x0 xe3 x0 x8 x0
lpfc 0000:0a:01.1: 1:0716 FCP Read Underrun, expected 255, residual 227
Data: xff xa3 x0


Thank your for your help!
Marius

[Michael Hockey]

Hi Marius
Im adding to the observations from a REdhat AS 4 box --- not at latest patch levels.

box 1
--------
/dev/sda:
Timing buffered disk reads: 18 MB in 3.02 seconds = 5.96 MB/sec

/dev/sda:
Timing buffered disk reads: 198 MB in 3.00 seconds = 65.97 MB/sec

box 2 (same MSA1500 different logical drives)
------
/dev/sda:
Timing buffered disk reads: 2 MB in 3.19 seconds = 642.51 kB/sec

/dev/sda:
Timing buffered disk reads: 152 MB in 4.41 seconds = 34.47 MB/sec

box 3 (same MSA1500 -- different logical drives)
---------
/dev/sda:
Timing buffered disk reads: 176 MB in 3.01 seconds = 58.42 MB/sec


Servers are Proliants -- typical CCIS respons e is
/dev/cciss/c0d0:
Timing buffered disk reads: 256 MB in 3.08 seconds = 83.05 MB/sec

So my response is jumping around , but can be very slow from the attached array.

re the lpfc errors, they are boot up messages only or do you see them in syslog on a booted system as well ?

The systems that I have with the MSA1500 are poor performing -- have great processors, lots of RAM but things run very slowly.