Fusion MPT driver version 3.01.20 VS. version 2.03.00
Hi
We have a bunch of IBM 335/336 used as email (anti-virus/spam) gateways. Recently our admin changed the kernel version from version 2.4.24 to 2.6.8.1 - this resulted in a massive performance degradation. The old system could handle 60.000 emails pr. hour the new one could only handle about 30.000 emails (this was tested by sampling 60.000 emails and sending them to the server).
After poking a bit around (changing hardware and upgrading bios's) I realized that the time spend for writing the files to the disk have gone up (multiple small writes seams to kill it): A hdparm test gives:
----------------------------
kernel 2.6.8.1 (MPT version : 3.01.09)
hdparm -t /dev/sda
/dev/sda:
Timing buffered disk reads: 64 MB in 3.03 seconds = 21.12 MB/sec
kernel 2.4.24 (MPT version: 2.03.00)
hdparm -t /dev/sda
/dev/sda:
Timing buffered disk reads: 64 MB in 1.19 seconds = 53.78 MB/sec
----------------------------
Next I tried upgrading the kernel to 2.6.12-rc5 to see if there was any improvements but there was none.
----------------------------
kernel 2.6.12-rc5 (MPT version : 3.01.20)
hdparm -t /dev/sda
/dev/sda:
Timing buffered disk reads: 64 MB in 3.03 seconds = 21.12 MB/sec
----------------------------
I have also tried building the driver from kernel 2.6.12-rc5 in the kernel instead of a module, this also changed nothing. The driver in the old kernel is 2.03.00 the new one is 3.01.20. On both kernel 2.6 and 2.4 /proc/scsi/scsi gives me
----------------------------
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: IBM-ESXS Model: CBR036C31210ESFN Rev: DFQ8
Type: Direct-Access ANSI SCSI revision: 04
Host: scsi0 Channel: 00 Id: 08 Lun: 00
Vendor: IBM Model: 25P3495a S320 1 Rev: 1
Type: Processor ANSI SCSI revision: 02
----------------------------
Regards.
Lars Roland
Quoted 11 lines · Lars Roland
We have a bunch of IBM 335/336 used as email (anti-virus/spam) gateways. Recently our admin changed the kernel version from version 2.4.24 to 2.6.8.1 - this resulted in a massive performance degradation. The old system could handle 60.000 emails pr. hour the new one could only handle about 30.000 emails (this was tested by sampling 60.000 emails and sending them to the server).
After poking a bit around (changing hardware and upgrading bios's) I realized that the time spend for writing the files to the disk have gone up (multiple small writes seams to kill it)
The driver was broken for a while in a way which caused it to run slowly if it was linked into the kernel image, but it runs OK if loaded as a module.
<digs around>
Here's a minimal patch - it fixes it, but iirc, this isn't what was merged. (Or you could just use 2.6.12-rc6).
drivers/message/fusion/mptscsih.c renamed from drivers/message/fusion/mptscsih.c~mpt-fusion-speedup +0 -4
@@ -96,7 +96,6 @@ MODULE_AUTHOR(MODULEAUTHOR);MODULE_DESCRIPTION (my_NAME);MODULE_LICENSE ("GPL" ); -#ifdef MODULE static int dv =MPTSCSIH_DOMAIN_VALIDATION ;module_param (dv, int,0 );MODULE_PARM_DESC (dv,"DV Algorithm: enhanced = 1, basic = 0 (default=MPTSCSIH_DOMAIN_VALIDATION=1)" );
@@ -112,7 +111,6 @@ MODULE_PARM_DESC(factor, "Min Sync Factostatic int saf_te =MPTSCSIH_SAF_TE ;module_param (saf_te, int,0 );MODULE_PARM_DESC (saf_te,"Force enabling SEP Processor: (default=MPTSCSIH_SAF_TE=0)" ); -#endif /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/
@@ -1477,7 +1475,6 @@ mptscsih_init(void) ": Registered for IOC reset notifications\n")); } -#ifdef MODULE dinitprintk((KERN_INFO MYNAM": Command Line Args: dv=%d max_width=%d " "factor=0x%x saf_te=%d\n",
@@ -1487,7 +1484,6 @@ mptscsih_init(void) driver_setup.max_width = (width) ?1 :0 ; driver_setup.min_sync_factor = factor; driver_setup.saf_te = (saf_te) ?1 :0 ;; -#endif if (mpt_device_driver_register (&mptscsih_driver,MPTSCSIH_DRIVER ) !=0 ) {
_
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 34 lines · Sonny Rao
Hi, I have an x335 here that I'm using on 2.6.11.11
<from dmesg>
Fusion MPT base driver 3.01.18
Copyright (c) 1999-2004 LSI Logic Corporation
ACPI: PCI interrupt 0000:01:01.0[A] -> GSI 22 (level, low) -> IRQ 22
mptbase: Initiating ioc0 bringup
ioc0: 53C1030: Capabilities={Initiator}
Fusion MPT SCSI Host driver 3.01.18
scsi0 : ioc0: LSI53C1030, FwRev=01000e00h, Ports=1, MaxQ=222, IRQ=22
Vendor: IBM-ESXS Model: MAS3367NC FN Rev: C901
Type: Direct-Access ANSI SCSI revision: 03
SCSI device sda: 71096640 512-byte hdwr sectors (36401 MB) SCSI device sda: drive cache: write through
SCSI device sda: 71096640 512-byte hdwr sectors (36401 MB) SCSI device sda: drive cache: write through
/dev/scsi/host0/bus0/target0/lun0: p1 p2
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Vendor: IBM-ESXS Model: MAS3367NC FN Rev: C901
Type: Direct-Access ANSI SCSI revision: 03
SCSI device sdb: 71096640 512-byte hdwr sectors (36401 MB) SCSI device sdb: drive cache: write through
SCSI device sdb: 71096640 512-byte hdwr sectors (36401 MB) SCSI device sdb: drive cache: write through
/dev/scsi/host0/bus0/target1/lun0: unknown partition table
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
Vendor: IBM Model: 25P3495a S320 1 Rev: 1
Type: Processor ANSI SCSI revision: 02Drive is a 15k 36GB disk:
hdparm -t
/dev/sda
Quoted 8 lines
/dev/sda:
Timing buffered disk reads: 226 MB in 3.01 seconds = 75.07 MB/secSo I'm not seeing that problem, does your dmesg output from the driver pretty much match what I have?
Sonny
Our is a 10K drive - this is my dmesg output (from 2.6.12-rc5) - only difference I can see is the ANSI SCSI revision (I have never investigated the SCSI subsystem much in the kernel so how does one interpreted the ANSI SCSI revision number ?).
Fusion MPT base driver 3.01.20
Copyright (c) 1999-2004 LSI Logic Corporation
ACPI: PCI Interrupt 0000:01:01.0[A] -> GSI 22 (level, low) -> IRQ 169
mptbase: Initiating ioc0 bringup
ioc0: 53C1030: Capabilities={Initiator}
Fusion MPT SCSI Host driver 3.01.20
scsi0 : ioc0: LSI53C1030, FwRev=01032715h, Ports=1, MaxQ=222, IRQ=169
Vendor: IBM-ESXS Model: CBR036C31210ESFN Rev: DFQ8
Type: Direct-Access ANSI SCSI revision: 04
SCSI device sda: 71096640 512-byte hdwr sectors (36401 MB) SCSI device sda: drive cache: write through
SCSI device sda: 71096640 512-byte hdwr sectors (36401 MB) SCSI device sda: drive cache: write through
/dev/scsi/host0/bus0/target0/lun0: p1 p2 p3 p4 < p5 p6 p7 >
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Vendor: IBM Model: 25P3495a S320 1 Rev: 1
Type: Processor ANSI SCSI revision: 02
Fusion MPT misc device (ioctl) driver 3.01.20
I have tried switching the 335 for another one but the performance is till very low so I am quite sure that it is not a hardware problem. Also compiling the driver as a module does not make any difference.
Regards.
Lars Roland
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 4 lines · Sonny Rao
Yeah, the only other thing I noticed is that your IRQ number is really large compared to mine, I'm not sure if that's some artifact of ACPI or what is going on there. Have you tried turning off ACPI or pci=routeirq and the like?
Nope but I will look at it.
Quoted 19 lines
My
/proc/interruptslooks like this:CPU0 CPU1 CPU2 CPU3
0: 80270452 0 0 0 IO-APIC-edge
timer1: 31529 0 0 0 IO-APIC-edge i8042 5: 0 0 0 0 IO-APIC-level acpi 8: 1 0 0 0 IO-APIC-edge rtc 11: 0 0 0 0 IO-APIC-level ohci_hcd 12: 503960 0 0 0 IO-APIC-edge i8042 15: 721836 0 0 0 IO-APIC-edge ide1 22: 152445 0 0 0 IO-APIC-level ioc0 25: 922998 0 0 0 IO-APIC-level eth1 NMI: 0 0 0 0 LOC: 80274015 80273964 80274024 80274023 ERR: 0 MIS: 0
Quoted 1 line
Is yours similar?
My looks like this on kernel 2.6:
---------------------
CPU0 CPU1 CPU2 CPU3
0: 13089 0 251098584 0 IO-APIC-edge timer 1: 8 0 0 0 IO-APIC-edge i8042 5: 0 0 0 0 IO-APIC-level acpi 8: 73 0 0 0 IO-APIC-edge rtc 11: 0 0 0 0 IO-APIC-level ohci_hcd:usb1 12: 110 0 0 0 IO-APIC-edge i8042 15: 12 0 0 0 IO-APIC-edge ide1 169: 24892566 0 0 0 IO-APIC-level ioc0 177: 121556391 0 0 0 IO-APIC-level eth0 NMI: 0 0 0 0 LOC: 251145010 251145083 251145081 251145081 ERR: 0 MIS: 0
---------------------
using kernel 2.4 it looks like this
---------------------
CPU0 CPU1 CPU2 CPU3
0: 128904101 0 0 0 IO-APIC-edge timer 1: 2 0 0 0 IO-APIC-edge keyboard 2: 0 0 0 0 XT-PIC cascade 8: 358 0 0 0 IO-APIC-edge rtc 11: 0 0 0 0 IO-APIC-level usb-ohci 15: 2 0 0 0 IO-APIC-edge ide1 22: 68048860 0 0 0 IO-APIC-level ioc0 24: 610278021 0 0 0 IO-APIC-level eth0 NMI: 0 0 0 0 LOC: 128905644 128905629 128905632 128905638 ERR: 0 MIS: 0
---------------------
I think you have good points about the irq beeing high.
Regards.
Lars Roland