SCSI pass through command cause iscsi Conn error

2009-09-28 Thread niko scsi 

I followed the howto http://tldp.org/HOWTO/SCSI-Generic-HOWTO/scsi_snd_cmd.html
to pass through scsi cmd with data (self defined,target can handle
it)  to iscsi disk.
 however,the request was handled six times ,I checked the log ,the
reason is shown below. The connection failed and recovered six
times ,and I tested many times ,each time ,the same :six times!
I'm newbie to scsi program. Any idea that can cause this problem ?

error occurred in function iscsi_data_rsp

tcp debug infomations show below :

...
Sep 27 09:37:01 localhost kernel: tcp: skb dfab65c0 ptr=da5a5a54
avail=60
Sep 27 09:37:01 localhost kernel: tcp: copied 0 0 size 48 recv
Sep 27 09:37:01 localhost kernel: tcp: iscsi_tcp_segment_recv copying
48
Sep 27 09:37:01 localhost kernel: tcp: copied 0 48 size 48 recv
Sep 27 09:37:01 localhost kernel: tcp: iscsi_tcp_segment_unmap
cb76be10
Sep 27 09:37:01 localhost kernel: tcp: total copied 48 total size 48
Sep 27 09:37:01 localhost kernel: tcp: segment done
Sep 27 09:37:01 localhost kernel: tcp: opcode 0x25 ahslen 0 datalen 12
Sep 27 09:37:01 localhost kernel: tcp: iscsi_data_rsp: ctask-
>exp_datasn(1) != rhdr->datasn(0)
Sep 27 09:37:01 localhost kernel: tcp: Error receiving PDU, errno=1001
Sep 27 09:37:01 localhost kernel:  connection1:0: detected conn error
(1011)
Sep 27 09:37:01 localhost kernel: tcp: iscsi_tcp_segment_unmap
cb76be10
Sep 27 09:37:01 localhost iscsid: Kernel reported iSCSI connection 1:0
error (1011) state (3)
Sep 27 09:37:03 localhost kernel: tcp: iscsi_tcp_hdr_recv_prep
(cb76be00)
Sep 27 09:37:03 localhost kernel: tcp: iscsi_tcp_send_hdr_prep
(cb76be00)
.

--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-iscsi@googlegroups.com
To unsubscribe from this group, send email to 
open-iscsi+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/open-iscsi
-~--~~~~--~~--~--~---

RE: connection error kills iscsi session

2009-09-28 Thread Drew Morris 

Mike,

Thanks for your prompt/detailed reply.

Vtrak is an iscsi disk array, which unfortunately doesn't seem to be reporting 
any errors.
Is there any sort of debugging I could do to discern what the cause of the 
problem is? I'm pretty sure it's not a pulled cable, but either the target or 
the initiator dropping the connection for some reason...
What sort of "invalid data" would cause the initiator to drop the connection?

There are no errors about pings or nops. The fragment I sent was the 
"beginning" of the problem. Can the verbosity of the iscsi messages be 
increased?

Below is the output of iscsiadm -m session -P 3. The connection failed again 
over the weekend, this time I wasn't even doing any reads/writes.


Some different error messages in there too, though I've seen these before.
Sep 27 02:51:04 spin kernel:  connection4:0: iscsi: detected conn error (1011)
Sep 27 02:51:05 spin iscsid: Kernel reported iSCSI connection 4:0 error (1011) s
tate (3)
Sep 27 02:51:08 spin iscsid: connect to 142.20.199.117:3260 failed (Connection r
efused)
Sep 27 02:51:41 spin last message repeated 9 times
Sep 27 02:52:45 spin last message repeated 17 times
Sep 27 02:53:49 spin last message repeated 17 times
Sep 27 02:54:52 spin last message repeated 17 times
Sep 27 02:55:56 spin last message repeated 17 times
Sep 27 02:56:00 spin iscsid: connect to 142.20.199.117:3260 failed (Connection r
efused)
Sep 27 02:56:04 spin kernel:  session4: iscsi: session recovery timed out after
300 secs
Sep 27 02:56:04 spin iscsid: connect to 142.20.199.117:3260 failed (Connection r
efused)
...
Sep 27 04:02:33 spin last message repeated 7 times
Sep 27 04:02:35 spin kernel: iscsi: cmd 0x28 is not queued (8)
Sep 27 04:02:35 spin kernel: sd 8:0:0:0: SCSI error: return code = 0x0001
Sep 27 04:02:35 spin kernel: end_request: I/O error, dev sde, sector 549316386
Sep 27 04:02:35 spin kernel: I/O error in filesystem ("sde1") meta-data dev sde1
 block 0x20bde700   ("xfs_trans_read_buf") error 5 buf count 8192
Sep 27 04:02:37 spin iscsid: connect to 142.20.199.117:3260 failed (Connection r
efused)
Sep 27 04:02:39 spin kernel: iscsi: cmd 0x28 is not queued (8)
Sep 27 04:02:39 spin kernel: sd 8:0:0:0: SCSI error: return code = 0x0001
Sep 27 04:02:39 spin kernel: end_request: I/O error, dev sde, sector 446314
Sep 27 04:02:39 spin kernel: I/O error in filesystem ("sde1") meta-data dev sde1
 block 0x6cf48   ("xfs_trans_read_buf") error 5 buf count 4096
...



iSCSI Transport Class version 2.0-724
version 2.0-871
Target: reserve.vtrak
Current Portal: 142.20.199.117:3260,1
Persistent Portal: 142.20.199.117:3260,1
**
Interface:
**
Iface Name: default
Iface Transport: tcp
Iface Initiatorname: unknown
Iface IPaddress: 142.20.199.72
Iface HWaddress: default
Iface Netdev: default
SID: 4
iSCSI Connection State: TRANSPORT WAIT
iSCSI Session State: Unknown
Internal iscsid Session State: REOPEN

Negotiated iSCSI params:

HeaderDigest: None
DataDigest: None
MaxRecvDataSegmentLength: 131072
MaxXmitDataSegmentLength: 524288
FirstBurstLength: 131072
MaxBurstLength: 262144
ImmediateData: Yes
InitialR2T: Yes
MaxOutstandingR2T: 1

Attached SCSI devices:

Host Number: 8  State: running
scsi8 Channel 00 Id 0 Lun: 0
Attached scsi disk sde  State: running

-Original Message-
From: Mike Christie [mailto:micha...@cs.wisc.edu]
Sent: Friday, September 25, 2009 8:46 PM
To: open-iscsi@googlegroups.com
Cc: Drew Morris
Subject: Re: connection error kills iscsi session

djmorris wrote:
> Problem: While running a backup to a mounted iscsi device, iscsid gets
> a "conn error (1011)", and my drive starts getting input/output
> errors.
>
> Details:
> Kernel Version: 2.6.23.17-88.fc7PAE
> Distribution: Fedora core 7
> open-iscsi version: 2.0-871
>
> ISCSI Target Device: Promise Vtrak M610i
> backup program: BackupPC (uses rsync)
> my linux box and target are both plugged into a Netgear 516T Switch
> (1000 baseT, 16 port, unmanaged)
>
> When I log into the target after this happens, i notice that there are
> no active iscsi sessions
> However, when i run
> iscsiadm -m session, it lists the active session
>
> tcp: [2] TAR.GET.IP.ADR:3260,1 reserve.vtrak
>
> I can log out  with
> iscsiadm -m node -U manual
> but i cannot log in after that or even discover the target device
> unless i reboot that device
>
> sample from my /var/log/messa

Intel NIC iSCSI acceleration in Linux and open-iscsi

2009-09-28 Thread Pasi Kärkkäinen 

Hello,

Is anyone working on Linux/open-iscsi iSCSI offloading/acceleration 
using Intel gigabit and 10 gigabit NICs? 

-- Pasi


--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-iscsi@googlegroups.com
To unsubscribe from this group, send email to 
open-iscsi+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/open-iscsi
-~--~~~~--~~--~--~---

Re: [PATCH v2 0/2] cxgb3/cxgb3i: added support of private MAC address and provisioning packet handler for iSCSI

2009-09-28 Thread David Miller 

From: "Karen Xie" 
Date: Fri, 25 Sep 2009 15:34:22 -0700

> Hmm, I am wondering how could this merge activity to be coordinated? If
> only the driver/scsi change is merged, then it won't compile either,
> since it requires the driver/net change.

That's rediculious, frankly.

Since they are two seperate changes you are knowingly creating
a bisection point that will not work.  That's wrong.

You need to split up the changes so that each and every one of them
are independant and the tree can be checked out at either of them and
everything can be expected to work.

--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-iscsi@googlegroups.com
To unsubscribe from this group, send email to 
open-iscsi+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/open-iscsi
-~--~~~~--~~--~--~---