Weird Qlogic 2200 FC HBA / CNSI G7324 problem

Weird Qlogic 2200 FC HBA / CNSI G7324 problem

[Thomas Langås]

We have an installation that constists of:
* 2x2 CNSI G7324 (active-active) 
* Each CNSI G7324 has 2x270GB RAID-partitions
* Servers with Qlogic 2200 FC HBAs

Today this is a 8 server, 1 HBA pr server config, so
there's one partition pr server. We're however trying to
change this now, to a 4 server, 2 HBA pr. server config
which will mean 2 partitions pr server. And now we bumped
into a problem; we can't get more than one partition
pr CNSI detected on each server, it doesn't matter how
many HBAs we insert, only one of the HBAs find a 
partition from that one CNSI-controller.

test3:/usr/src/linux-2.4/drivers/scsi# uname -a
Linux test3.stud.ntnu.no 2.4.18-3smp #1 SMP Thu Apr 18 07:27:31 EDT 2002 i686 unknown

(Output from dmesg)
scsi1 : QLogic QLA2200 PCI to Fibre Channel Host Adapter: bus 1 device 6 irq 16
        Firmware version:  2.02.03, Driver version 6.1b2
scsi2 : QLogic QLA2200 PCI to Fibre Channel Host Adapter: bus 1 device 8 irq 20
        Firmware version:  2.02.03, Driver version 6.1b2

Anyone has any idea where we should start searching for
a solution to this problem?  


Another problem, which I don't think is serious, is this:

(also from dmesg)

scsi1: Host table full.
scsi2: Host table full.

I get this on all our newly installed servers, never seen 
this one before, actually. But looking through the driver
code, it didn't seem that it was something to care about:

test3:/usr/src/qlogic# grep -B 18 -A 13 "Host table" qla2x00.c 
                } else if (status[0] == 0x4000) {
                        if (status[1] & 0x0001) {
                                /*
                                 * This is an initiator only device. Save
                                 * some info to host_db and ignore it.
                                 */
                                if (ha->next_host_slot < 8) {
                                        BCOPY(device->wwn, ha->phost_db
                                            [ha->next_host_slot].wwn, 8);
                                        BCOPY(device->name, ha->phost_db
                                            [ha->next_host_slot].name, 8);
                                        ha->phost_db[ha->next_host_slot].
                                            d_id.b24 = device->d_id.b24;
                                        ha->phost_db[ha->next_host_slot].
                                            loop_id = device->loop_id;
                                        ha->next_host_slot++;
                                } else {
                                        printk(KERN_INFO
                                            "scsi%ld: Host table full.\n",
                                            ha->host_no);
                                }
                                return 2;
                        }

                        /* This is target capable device */
                        qla2x00_get_port_database(ha, device, 0);

                        DEBUG(printk("scsi%d: Fabric Login OK. loop id=0x%04x, "
                            "port Id=%06x\n",
                            (int)ha->host_no, 
                            device->loop_id, device->d_id.b24);)
                        return 0;


So, any help at all would be greatly appreciated.
In advance, thanks!

-- 
Thomas

Re: Weird Qlogic 2200 FC HBA / CNSI G7324 problem

[Thomas Langås]

Mike Anderson:
> Can you send the output of "/proc/scsi/qla2200/1" and "/proc/scsi/qla2200/2" 

test3:~# cat /proc/scsi/qla2200/1
QLogic PCI to Fibre Channel Host Adapter for ISP22xx/ISP23xx:
        Firmware version:  2.02.03, Driver version 6.1b2
Entry address = f883a060
HBA: QLA2200 , Serial# B50409
Request Queue = 0x32d58000, Response Queue = 0x32d50000
Request Queue count= 128, Response Queue count= 64
Total number of active commands = 0
Total number of interrupts = 352
Total number of IOCBs (used/max) = (0/600)
Total number of queued commands = 0
    Device queue depth = 0x10
Number of free request entries = 116
Number of mailbox timeouts = 0
Number of ISP aborts = 0
Number of loop resyncs = 3
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= <READY>, flags= 0x8a0813
Dpc flags = 0x0
MBX flags = 0x0
SRB Free Count = 4096
Port down retry = 008


SCSI Device Information:
scsi-qla0-adapter-node=200000e08b02894b;
scsi-qla0-adapter-port=210000e08b02894b;
scsi-qla0-target-0=2000005013d01808;
scsi-qla0-target-1=2000005013d016d3;
scsi-qla0-target-2=2000005013d0158d;
scsi-qla0-target-3=2000005013d0154b;

SCSI LUN Information:
(Id:Lun)
( 0: 0): Total reqs 7, Pending reqs 0, flags 0x0, 0:0:82,
test3:~# cat /proc/scsi/qla2200/2
QLogic PCI to Fibre Channel Host Adapter for ISP22xx/ISP23xx:
        Firmware version:  2.02.03, Driver version 6.1b2
Entry address = f883a060
HBA: QLA2200 , Serial# B50405
Request Queue = 0x32d38000, Response Queue = 0x33060000
Request Queue count= 128, Response Queue count= 64
Total number of active commands = 0
Total number of interrupts = 10
Total number of IOCBs (used/max) = (0/600)
Total number of queued commands = 0
    Device queue depth = 0x10
Number of free request entries = 117
Number of mailbox timeouts = 0
Number of ISP aborts = 0
Number of loop resyncs = 3
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= <READY>, flags= 0x8a0813
Dpc flags = 0x0
MBX flags = 0x0
SRB Free Count = 4096
Port down retry = 008


SCSI Device Information:
scsi-qla1-adapter-node=200000e08b02854b;
scsi-qla1-adapter-port=210000e08b02854b;
scsi-qla1-target-0=2000005013d01808;
scsi-qla1-target-1=2000005013d016d3;
scsi-qla1-target-2=2000005013d0158d;
scsi-qla1-target-3=2000005013d0154b;

SCSI LUN Information:
(Id:Lun)
( 2: 0): Total reqs 7, Pending reqs 0, flags 0x0, 1:0:86,



-- 
Thomas

Re: Weird Qlogic 2200 FC HBA / CNSI G7324 problem

[Thomas Langås]

Just a quick update; doing the following works:

echo "scsi add-single-device 1 0 0 1" > /proc/scsi/scsi


Which indicates that's the detection code somewhere which
isn't like it should?

-- 
Thomas

Re: Weird Qlogic 2200 FC HBA / CNSI G7324 problem

[Thomas Langås]

Thomas Langås:
> Just a quick update; doing the following works:
> echo "scsi add-single-device 1 0 0 1" > /proc/scsi/scsi
> Which indicates that's the detection code somewhere which
> isn't like it should?


And, to keep on replying to myself; the redhat kernel didn't 
have "Probe all SCSI LUNs" enabled, adding that and it finds
everything like it should. Any reason why it doesn't without
this option?

-- 
Thomas
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Weird Qlogic 2200 FC HBA / CNSI G7324 problem

[Mike Anderson]

Thomas Langås [tlan@stud.ntnu.no] wrote:
> Thomas Langås:
> > Just a quick update; doing the following works:
> > echo "scsi add-single-device 1 0 0 1" > /proc/scsi/scsi
> > Which indicates that's the detection code somewhere which
> > isn't like it should?
> 
> 
> And, to keep on replying to myself; the redhat kernel didn't 
> have "Probe all SCSI LUNs" enabled, adding that and it finds
> everything like it should. Any reason why it doesn't without
> this option?

With this config option not selected non zero luns will not be probed. You
can also pass max_scsi_luns on the command line without re-compiling your
kernel. Someone from RedHat would be better to comment on the default
selection, but as you will see if you review the archive of this mail list
there have been passed issues of probing non-zero luns on older devices
not included in the scsi devicelist, so defaulting to off is possibly a
conservative selection.

-Mike
-- 
Michael Anderson
andmike@us.ibm.com

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Weird Qlogic 2200 FC HBA / CNSI G7324 problem

[Thomas Langås]

Mike Anderson:
> With this config option not selected non zero luns will not be probed. You
> can also pass max_scsi_luns on the command line without re-compiling your
> kernel. Someone from RedHat would be better to comment on the default
> selection, but as you will see if you review the archive of this mail list
> there have been passed issues of probing non-zero luns on older devices
> not included in the scsi devicelist, so defaulting to off is possibly a
> conservative selection.

Tried the max_scsi_luns=128 (which is max, right?) without luck, before
I recompiled the kernel with "Probe all SCSI LUNs" enabled.

-- 
Thomas