[EMAIL PROTECTED] wrote:
Hi experts,
I have some questions about LinkDown of SG-XPCI2FC-QF2-Z.
I got the following advice from Mr. Richard.
Please send me any comments or advices.
Best regards,
Satoshi Yajima
-------- Original Message --------
Subject: Re: Second Try: about LinkDown of SG-XPCI2FC-QF2-Z.
Date: Wed, 30 Jul 2008 22:44:39 -0700
From: Richard Elling <[EMAIL PROTECTED]>
[EMAIL PROTECTED] wrote:
Sorry to send this mail again.
I'm not sure there are many fibre channel experts on this alias.
You might also ask on the external [email protected]
mail alias.
My impressions below...
I have a customer who is waiting for an immediate answer of the
following question.
But, I have not received the information necessary for the answer.
Please send me any comments or informations.
Thanks and Regards,
Satoshi Yajima
-------- Original Message --------
Subject: about LinkDown of SG-XPCI2FC-QF2-Z.
Date: Fri, 25 Jul 2008 13:06:09 +0900
From: [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
CC: Mukaeda Toshiki <[EMAIL PROTECTED]>, Satoshi Yajima
<[EMAIL PROTECTED]>
Hi experts,
These are four questions concerning the message at the following
configuration.
Server : Sun Fire V440
OS : Solaris 8 2/04
HBA : SG-XPCI2FC-QF2-Z
1.When link-down occurs in the above HBA, does the message output
to Syslog ?
What is the log level ?
(for example kern.info etc.)
Probably not. The reason is that a switched fabric and arbitrated loop
handle these conditions differently. Most HBAs are designed to work
for both. In an arbitrated loop configuration, a loop reconfiguration
may occur relatively frequently, but while it is detected by all nodes
on the loop, it is not considered a failure condition.
2. If a switch or a storage connects to above configuration,and
when link-down occurs in the switch or the storage,does the HBA
go link-down and notify ?
If the storage is connected with the host directly, when the storage's
link is down, the HBA will go link-down.
If they are connected through the switch,
when the storage goes link down, the HBA will not go link-down, but it
will get notified, and the switch will send RSCN to the HBA.
When the switch goes goes down, the HBA will go link-down, too.
Regards,
Reed
3. Even when the error occurs only in one of RX or TX, does the
message output ?
Or only when the error occurs in both,does the message output as
link-down ?
AFAIK, TX errors are not detected in fibre channel. The HBA is
constantly transmitting and link detection can only occur on the
receiving side.
4.When link-down occurs between storage to the above environment,and
if it is failed to write the storage,does the message output to syslog ?
What is the log level ?
(for example kern.info etc.)
Device timeouts which occur at the sd or ssd driver occur after
60 second retry intervals with 3-5 retries. For most fibre channel
interfaces, there will be 3 retries. For each retry, a syslog message
is recorded with the indication that a retry is in progress. After the
timeout and retry intervals are exceeded, then the errors will
propagate up the software stack.
For Solaris NV b72 or later, the FMA I/O retire diagnosis engine
will report diagnosis and can be configured to provide SNMP
traps. However, usually the failure will be noted by the higher-
level software like file systems.
-- richard
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
|
