[OpenIndiana-discuss] problems with iscsi/taget going offline

[Jonathan Loran]

Hi List,

I've been fighting an intermittent bugaboo that's driving me crazy.  I haven't found anything on the list that shows anyone seeing anything like this.  I have a Solaris 10 server that mounts a zpool off of an Openindiana storage server using disks served by iscsi (Comstar).  Every so often, sometimes every day, then not for weeks the iscsi target just goes south on the Openindiana box.  There are scant messages on the Openindiana iscsi target, and the iscsi/target service claims to stay online.  But on the console on the Sol 10 box, there are messages like:

...
 Aug  1 18:00:29 core-bu scsi: WARNING: /scsi_vhci/disk at g600144f0f7c5880000004fc7a1df0015 (sd55):
 Aug  1 18:00:29 core-bu         SCSI transport failed: reason 'tran_err': retrying command
 Aug  1 18:00:52 core-bu iscsi: NOTICE: iscsi connection(11) unable to connect to target iqn.2007-07.edu.berkeley.ssl:iscsi-9.target0
 Aug  1 18:00:53 core-bu iscsi: NOTICE: iscsi connection(7) unable to connect to target iqn.2007-07.edu.berkeley.ssl:iscsi-9.target0
 Aug  1 18:02:00 core-bu scsi: WARNING: /scsi_vhci/disk at g600144f0f7c5880000004fc7a1df0014 (sd54):
 Aug  1 18:02:00 core-bu         SCSI transport failed: reason 'timeout': retrying command
 Aug  1 18:02:01 core-bu scsi: WARNING: /scsi_vhci/disk at g600144f0f7c5880000004fc7a1df000b (sd45):
...

and naturally the Sol 10 box hangs.  The Solaris 10 server is used for various backups, and is thus hammered pretty hard at times.  However, it seems like the failures are not load related, since they often happen at off-peak times.

Has anyone seen a problem like this?  I'm not sure which end is the problem here: the Openindiana side or Sol 10.  Here are the versions:

Sol 10 box:
# cat /etc/release
                    Oracle Solaris 10 9/10 s10x_u9wos_14a X86
     Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
                            Assembled 11 August 2010

Openindiana box (iscsi server):
# cat /etc/release 
                      OpenIndiana Development oi_151a X86
        Copyright 2010 Oracle and/or its affiliates. All rights reserved.
                        Use is subject to license terms.
                           Assembled 01 September 2011

Thanks,

Jon



-     _____/     _____/      /           - Jonathan Loran -           -
-    /          /           /                IT Officer               -
-  _____  /   _____  /     /     Space Sciences Laboratory, UC Berkeley
-        /          /     /                (510) 643-5146             -
- ______/    ______/    ______/        jloran at ssl.berkeley.edu        -