SUMMARY: JNI via Fabric

From: <jhartzen_at_csc.com>
Date: Mon Dec 10 2001 - 06:10:05 EST
Thank you to Fredrik Robertsson for a quick and accurate assessment of my
problem (minutes after I sent the question).  His response gets to go into
the Summaries archive :-).  Other people have suggested similar solutions
and sent me some documents with detailed settings.  Thanx to all of you.
To the person who checked that I had plugged the fibres into a switch, yes
I remembered to change them even though I didn't mention it <grin>

Hi!

If you read the comments in fcaw.conf carefully, you should see that
setting
public_loop overrides fca_nport. So set fca_nport = 1 and make sure that
public_loop = 0 (default), and no other changes should be needed.
Depending,
ofcourse, on your switches...

/frero


My original query below.

I would still appreciate any answers re the difference between
def_wwnn_binding and def_wwpn_binding.  I can see from things like luxadm
or dmesg both the port-name and the node-name... but how do I know when to
select which?

Thanx,
  _Johan

P.S. I accidentally had a "Re" in my Summary subject line, so it didn't go
through automatically





                                                                                                                                  
                    Johan                                                                                                         
                    Hartzenberg/G        To:     sunmanagers@sunmanagers.org                                                      
                    IS/CSC               cc:                                                                                      
                    @CSC                 Subject:     JNI via Fabric                                                              
                    Sent by:                                                                                                      
                    sunmanagers-a                                                                                                 
                    dmin                                                                                                          
                                                                                                                                  
                                                                                                                                  
                    07/12/2001                                                                                                    
                    10:51 AM                                                                                                      
                                                                                                                                  
                                                                                                                                  




Hi fellow guru's,

I've got some E10K domains connected to a Hitachi Lightning 9960.  Due to
the price of ports on the Lightning, we want to convert the FC connection
from direct point-to-point connections to switched "fabric-mode"
connections.

Can anyone tell me what settings need to change to do this conversion?  I
attempted the following, but it failed, maybe someone can point out what I
did wrong:


/kernel/drv/fcaw.conf

Changed the following settings:

fca_nport = 0 changed to 1;
public_loop = changed to 1;

I also played around with several combinations of the below

class2_enable = 1; (default off)
link_recovery_delay = 500; (increased the value)
timeout_reset_enable = 1; (default off)

I did not have time to play around with the heartbeat setting due to
change-slot time constraints, and was forced to regress the change in order
to bring the system back up.  This is the NetBackup Master, so changes
happen during daytime when it is mostly idle.

In any case, no matter what I tried, Solaris refused to recognise the FC
link.  The messages on the console indicates LINK DOWN (trimmed for
brevity)


Dec  5 12:13:17 zactcsui306 syslogd: going down on signal 15
stopping the Media Manager device daemon
stopping the Media Manager volume daemon
The system is down.
syncing file systems... done
[snip]
/kernel/drv/sparcv9/fcaw symbol ddi_model_convert_from multiply defined^A^V
fcaw0: JNI Fibre Channel Adapter model FCW
fcaw0: 64-bit SBus 0: IRQ 3: FCODE Version 13.3.7 [18c932]: SCSI ID 125:
AL_PA 01
fcaw0: Fibre Channel WWNN: 100000E0694152B6  WWPN: 200000E0694152B6
fcaw0: FCA SCSI/IP Driver Version 2.5.8.HIT.07, July 31, 2000 for 64bit
Solaris
fcaw0: All Rights Reserved.
fcaw0: Resetting GLM...
fcaw0: Link Failure. Resetting...

fcaw0: LINK DOWN: Check Connections...
fcaw1: JNI Fibre Channel Adapter model FCW
fcaw1: 64-bit SBus 1: IRQ 3: FCODE Version 13.3.7 [18c932]: SCSI ID 125:
AL_PA 01
fcaw1: Fibre Channel WWNN: 100000E0694152B6  WWPN: 200000E069415C7A
fcaw1: FCA SCSI/IP Driver Version 2.5.8.HIT.07, July 31, 2000 for 64bit
Solaris
fcaw1: All Rights Reserved.
fcaw1: Resetting GLM...
fcaw0: Link Failure. Resetting...

fcaw1: Link Failure. Resetting...
fcaw0: Link Failure. Resetting...
fcaw1: Link Failure. Resetting...
[snip]
fcaw1: LINK DOWN: Check Connections...
fcaw1: Link Failure. Resetting...
[snip]
fcaw0: Link Failure. Resetting...
WARNING: forceload of drv/atf failed
fcaw1: Link Failure. Resetting...
[snip]
fcaw0: Link Failure. Resetting...
configuring IPv4 interfaces: hme0 hme1.
Hostname: zactcsui306
fcaw1: Link Failure. Resetting...
fcaw0: Link Failure. Resetting...
vxvm: Volume s01 for mountpoint - is not in rootdg disk group;
        Cannot start the volume early
VxVM starting special volumes ( swapvol var var s01 )...
fcaw1: Link Failure. Resetting...
fcaw0: Link Failure. Resetting...
fcaw1: Link Failure. Resetting...
fcaw0: Link Failure. Resetting...
VxVM general startup...
      Dump content: kernel pages
       Dump device: /dev/dsk/c0t0d0s1 (dedicated)
Savecore directory: /var/crash/zactcsui306
  Savecore enabled: yes
fcaw1: Link Failure. Resetting...

fcaw0: Link Failure. Resetting...
The system is coming up.  Please wait.
fcaw1: Link Failure. Resetting...
Cannot stat /dev/vx/rdsk/hit306dg/d02 - ignoring
fcaw0: Link Failure. Resetting...
Cannot stat /dev/vx/rdsk/hit306dg/d01 - ignoring
Cannot stat /dev/vx/rdsk/hit306dg/tmp01 - ignoring
Dec  5 12:18:06 su: 'su sys' succeeded for root on /dev/console
fcaw1: Link Failure. Resetting...
starting rpc services:fcaw0: Link Failure. Resetting...
 rpcbind done.
Setting netmask of hme0 to 255.255.255.0
Setting netmask of hme1 to 255.255.252.0
Setting default IPv4 interface for multicast: add net 224.0/4: gateway
zactcsui306
/dev/vx/dsk/hit306dg/s01: No such file or directory
syslog service starting.
syslogd: line 33: WARNING: zactcsuc300 could not be resolved
/etc/rc2.d/S77netbackup: /usr/openv/volmgr/bin/vmps: not found
Media Manager daemons not started.
NetBackup request daemon not started.
Dec  5 12:18:10 zactcsui306 fcaw: fcaw1: Link Failure. Resetting...
Dec  5 12:18:10 zactcsui306 fcaw: fcaw0: Link Failure. Resetting...
Starting RMI Registry
Starting VERITAS VM Storage Administrator Command Server
Dec  5 12:18:12 zactcsui306 xntpd[290]: couldn't resolve `zactcsuc300',
giving up on it
Dec  5 12:18:13 zactcsui306 fcaw: fcaw1: Link Failure. Resetting...
Dec  5 12:18:13 zactcsui306 fcaw: fcaw0: Link Failure. Resetting...
Starting VERITAS VM Storage Administrator Server
awservices started 379
Dec  5 12:18:15 zactcsui306 fcaw: fcaw1: Link Failure. Resetting...
Dec  5 12:18:15 zactcsui306 fcaw: fcaw0: Link Failure. Resetting...
RUNNING    aws_orb:aws_orb
Dec  5 12:18:18 zactcsui306 fcaw: fcaw1: Link Failure. Resetting...
etc




Below is the settings I thought should work:
====================================
scsi-initiator-id = 0x7d;
fca_nport = 1;
public_loop = 1;
target_controllers = 126;
ip_disable = 1;
qfull_retry_count = 0;
qfull_retry_interval = 1000;
failover = 60;
failover_extension = 0;
recovery_attempts = 5;
class2_enable = 1;
fca_heartbeat = 0;
reset_glm = 1;
timeout_reset_enable = 1;
busy_retry_delay = 500;
link_recovery_delay = 500;
scsi_probe_delay = 0;
def_hba_binding  = "fcaw*";
def_wwpn_binding  = "$xxxxxxxxxxxxxxxx";
def_wwnn_binding  = "xxxxxxxxxxxxxxxx";
def_port_binding = "$xxxxxx";
fca_verbose = 1;

In addition, I don't expect any hassles, but do I need to worry about
Veritas Volume Manager not recognizing or automatically importing the disk
groups and volumes?  Could someone explain to me the difference between
PortName binding and Node-name binding in a way that would explain to me
how to decide which one to use.

Also very helpful would be a way to change fcaw settings without needing to
reboot, though I'm not holding my breath here.

Thanx,
  _Johan


P.S. If I can't get this to work soon I will look for a different HBA
manufacturer with better support.
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
Received on Mon Dec 10 11:10:05 2001

This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:32:37 EDT