Software
Interface Driver Global Interface Statistics | | |
Number of connection opens | I | The number
of TCP connection opens initiated.
The statistic is incremented when a call to open a connection is
made. |
Number of connection closes | I | The
number of TCP connection
closes initiated. The statistic is incremented when a call to close
a connection is made. |
Software Interface Driver Connection
Statistics | | |
Number of times the login failed | I/T | The number of login failures
due to incorrect text / key values and formatting. If this is a transient
problem at the target end, the initiator would recover on a successive
retry attempt. |
Number of exception status class values
returned by target | D | The number of iSCSI
login phase failure. The reasons for login failure can be determined
by looking at the STM/syslog.log logs to find more detailed information.
The class of login failures enumerated here consists of interoperability
(potential protocol violations) issues between the host and the
target. |
Number of PDU headers with Protocol errors
received by initiator | T | The number
of login failures due to protocol violations by the target. The
protocol violation occurs when a target sends a PDU login response
header and the initiator determines that the response is not protocol compliant. |
Number of times iswd daemon failed to
open a connection | D/C | The number of times the iswd daemon failed to open a connection to the requested
target. The failure can be the result of resource allocation failures,
incorrect target configuration, or network infrastructure problems.The
exact reason for the failed will be logged in syslog.log. |
Number of failures to send a login command
due to kernel memory allocation failure | D | The number of
attempts to send the Login command that failed due to memory allocation failures.
The upper level driver recovery may retry the session open, resulting in
a re-attempt to send the Login command. If the memory allocation
request succeeds, the Login command will transmit successfully. |
Number of asynchronous failures waiting
for a login response | D | The number of
asynchronous failures experienced while waiting for a login response.
The asynchronous failure might be due to a PDU exchange timeout/abort
or lack of memory resources. The upper level driver recovery may
retry the session open, resulting in a re-attempt to send the Login
command. If the memory allocation request succeeds, the Login command
will transmit successfully. |
Number of asynchronous failures waiting
for a logout response | D/C | The number of asynchronous failures experienced
while waiting for a logout response. The asynchronous failure might
be due to a PDU exchange timeout/abort or lack of memory resources.
One additional attempt to complete the Logout is made by requesting the
iswd daemon to close the TCP connection and tear down the stream. |
Number of unexpected TCP closes in the
active state | C | The number of
unexpected TCP close events received during the connection ready
state. As part of recovery, all the I/Os on this connection are
aborted, the connection is closed, and a session reopen is triggered. |
Number of timeouts on FIN after sending
a logout command | C | The number of timeouts on FIN after the target
has sent a logout response. A close is triggered by requesting
the iswd daemon to close the TCP connection and tear down the stream. |
Number of TCP connection open timeouts | C | The number of TCP connection open timeout occurs.
The timeout will trigger the freeing of resources. The upper level
driver recovery may retry the session open, resulting in a re-attempt
to send the Login command. If the memory allocation request succeeds,
the Login command will transmit successfully. |
Number of unexpected connection closes
after a login command | C | The number of
unexpected TCP close events received while waiting for a login response.
As part of the recovery mechanism, resources are freed, the connection
is closed, and a session reopen is triggered. |
Number of unexpected connection closes
after a logout command | C | The number of unexpected TCP close
events received while waiting for a logout response. As part of
the recovery mechanism, the Logout PDU is aborted, resources are
freed, and the TCP connection is closed. |
Number of target authentication timeouts | I/C | The number target authentication
timeouts that occurred during communication with the userspace iradd daemon. |
Number of target authentication failures | I/C | The number of target
authentication failures for CHAP. Either the target’s CHAP information
is not configured in the RADIUS server, or the CHAP information
provided by the target is incorrect. |
Number of temporary redirection requests | I | Number of temporary login redirections requested
by a target device. |
Number of permanent redirection requests | I | Number of permanent login redirections requested
by a target device. |
Number of kernel memory allocation failures | D | The number of memory allocation
attempts for a kernel PDU structure
that failed. A failure to allocate a PDU structure means that some
outbound command, or a NOP-OUT in response to a NOP-IN, could not
be completed. As a result: The regular occurrence
of this event will have a negative impact on performance. Failed
I/Os will be retried according to existing SCSI retry policies (same
as Fibre Channel and parallel SCSI), however, there is no retry
policy for native iSCSI commands.
|
Number of streams message allocation
failures | D | The number of streams message
memory allocation attempts that failed. The regular occurrence
of this event will have a negative impact on performance. Failed
I/Os will be retried according to existing SCSI retry policies (same
as Fibre Channel and parallel SCSI). |
Number of PDU transmission failures due
to an offline connection | D | The number of transmission attempts of native
iSCSI commands that failed, because the connection on which the
I/O was attempted, was offline. The regular occurrence of this
event will have a negative impact on performance. Failed I/Os will
be retried according to existing SCSI retry policies (same as Fibre
Channel and parallel SCSI). |
Number of streams message duplication
failures | D | The number of streams message duplication operations
that failed. As a result of the failure, the iSCSI Software Initiator
will not transmit the related iSCSI commands. The regular occurrence
of this event will have a negative impact on performance. Failed
I/Os will be retried according to existing SCSI retry policies (same
as Fibre Channel and parallel SCSI). |
Number of PDU exchange timeouts | D/C/T | The number of iSCSI PDUs successfully transmitted
to the target for which no response was received within a specified period
of time. If a timeout occurs, the iSCSI Software Initiator will
initiate a logout for the session as part of the recovery, and eventually will
attempt to login again with the target. The problem could be a
network infrastructure problem or a target congestion issue. The
regular occurrence of this event will have a negative impact on
performance. Failed I/Os will be retried according to existing SCSI
retry policies (same as Fibre Channel and parallel SCSI). |
Number of PDU exchanges aborted | C/T | The number of iSCSI PDUs that were aborted by
the initiator while waiting for a response from a target. The abort
occurs if a response to the iSCSI PDU has not been received at the initiator
within a preset timeout period. The failure can also be due to
an unexpected close of the TCP connection. The problem could be a
network infrastructure problem or a target congestion issue. |
Number of PDU exchanges abandoned | D/T | The number of exchanges between an iSCSI initiator
and target that were abandoned. This will typically occur when
the number of exchanges to complete a negotiation goes beyond a
predetermined limit, usually indicating an infinite loop. |
Number of I/Os issued on this connection | I | The number of SCSI
I/Os that were sent to the networking stack by the iSCSI Software Initiator. |
Number of I/O timeouts | D/T | The number of SCSI I/Os that
did not complete within a time period preset by an upper level protocol
(SCSI). The driver will recover from SCSI I/O timeouts using session level
error recovery, for example, tearing down the session and starting
over. Failed I/Os will be retried according to existing SCSI retry
policies (same as Fibre Channel and parallel SCSI). |
Number of kernel memory allocation failures | D | The number of kernel
memory allocation failures for SCSI I/O related data structures.
Failure to allocate will result in the I/O not being processed
and returned to the SCSI layer. The regular occurrence of this
event will have a negative impact on performance. Failed I/Os will
be retried according to existing SCSI retry policies (same as Fibre Channel
and parallel SCSI). |
Number of streams message allocation
failures | D | The number of memory allocation
attempts for a kernel driver structure that failed. A failure to
allocate the driver SCSI structure means that some SCSI I/O could
not be completed. The regular occurrence of this event will have
a negative impact on performance. Failed I/Os will be retried according
to existing SCSI retry policies (same as Fibre Channel and parallel
SCSI). |
Number of I/Os delayed while waiting
for resources | D | The number of
memory allocation attempts for a kernel driver structure that failed.
The iSCSI Software Initiator will retry the allocation request
a set number of times. The ultimate failure to allocate the driver structure
means that some SCSI I/O could not be completed. The regular occurrence
of this event will have a negative impact on performance. Failed
I/Os will be retried according to existing SCSI retry policies (same
as Fibre Channel and parallel SCSI). |
Number of I/Os failed due to an offline
connection | D | The number of SCSI I/Os that were aborted
as a result of retrying memory resource allocation when the connection
went offline. As a result of this event, the corresponding SCSI
I/O will be aborted. This could be a network infrastructure problem.
Failed I/Os will be retried according to existing SCSI retry policies
(same as Fibre Channel and parallel SCSI). |
Number of I/Os failed due to memory resource
constraints | D | The number of
SCSI I/Os that failed to acquire memory resources within the maximum
number of retries. As a result of the memory allocation failures,
the SCSI I/O was failed backed to SCSI. Failed I/Os will be retried
according to existing SCSI retry policies (same as Fibre Channel
and parallel SCSI). |
Number of invalid Data-In PDUs received | T | The total number
of invalid Data-In PDUs sent by the target. A non-zero value for
this stat indicates that the target is not adhering to the iSCSI
protocol by not sending Data-In PDUs with: Buffer Offset
in increasing offset order with non-overlapping ranges. DataSN
in increasing order. |
Number of I/O underruns | I | The number of I/Os
on a connection where the number of bytes that were received by
the initiator did not match the number of bytes that the target
sent. |
Number of I/O underflows | I | The number of I/Os on a connection where
the target sent less data than what was requested by the initiator.
A large value for this statistic is normal. |
Number of I/O overflows | I | The total number of I/Os on a connection where
the target sent more data than what was requested by the initiator.
This indicates that the Expected Data Transfer Length in the I/O
request was not sufficient. |
Number of I/O failures due to response
code errors | T | The total number of I/Os that failed
due to a response code error in the SCSI response PDU. This means
that the target failed to execute the I/Os. A large value here
could indicate that the target is not functioning properly. |
Number of Data-In PDUs received without
data | I | The total number
of empty Data-In PDUs received by the initiator. |
Number of invalid R2T PDUs received | T | The total number of Request to Transfer (R2T)
PDUs from the target that had an incorrect buffer offset or Data
Transfer Length in them. A non-zero value for this stat indicates
that the target is not adhering to the error recovery policy negotiated
by the initiator. |
Number of I/Os that failed to respond
to an R2T due to kernel memory allocation failures | D | The total number of Request to Transfer (R2T) PDUs that
could not be honored by the initiator due to resource constraints
at the initiator. A non-zero value for this stat indicates that
the initiator is intermittently running out of resources while handling
iSCSI I/O traffic. |
Number of unexpected R2T PDUs received
during a Read I/O | T | The number of Request to Transfer (R2T) requests
sent by the target for a Read operation. This is an unexpected behavior. |
Number of unexpected Data-In PDUs received
during a Write I/O | T | The total number of data-in
PDUs received by the initiator while the initiator was doing a write
operation to the target. This is an unexpected behavior. |
Number of Data-In PDUs with incorrect
residual count | T | The total number of I/O requests where
the target had an incorrect residual count value and where the status
for the I/O was sent as part of the last Data-In PDU. This might happen
when the target indicated an underflow condition but the residual
count value did not match the expected residual count. |
Number of SCSI Response PDUs with incorrect
residual count | T | The total number of I/O requests
where the target had an incorrect residual count value set. This
might happen when the target indicated an underflow condition but
the residual count value did not match the expected residual count.
The response PDU in this case was sent as a separate PDU by the target. |
Number of I/O underruns | I | The number of I/Os
on a connection where the number of bytes that were received by
the initiator did not match the number of bytes that the target
sent. |
Number of I/O underflows | I | The number of I/Os on a connection where
the target sent less data than what was requested by the initiator.
A large value for this statistic is normal. |
Number of I/O overflows | I | The total number of I/Os on a connection where
the target sent more data than what was requested by the initiator.
This indicates that the Expected Data Transfer Length in the I/O
request was not sufficient. |
Number of I/O failures due to response
code errors | T | The total number of I/Os that failed
due to a response
code error in the SCSI response PDU. This means
that the target failed to execute the I/Os. A large value here
could indicate that the target is not functioning properly. |
Number of Data-In PDUs received without
data | I | The total number of empty Data-In PDUs received
by the initiator. |
Number of invalid R2T PDUs received | T | The total
number of Request to Transfer (R2T) PDUs from the target that
had an incorrect buffer offset or Data Transfer Length in them.
A non-zero value for this stat indicates that the target is not
adhering to the error recovery policy negotiated by the initiator. |
Number of I/Os that failed to respond
to an R2T due to kernel memory allocation failures | D | The total number of Request
to Transfer (R2T) PDUs that could not be honored by the initiator
due to resource constraints at the initiator. A non-zero value
for this stat indicates that the initiator is intermittently running
out of resources while handling iSCSI I/O traffic. |
Number of I/O failures due to streams
message concatenation memory allocation failures | D | The total number of I/O failures
in the inbound path resulting from a failure of msgpullup call. |
Number of holes seen in the status sequencing | D/T | The total number of PDUs that
were received where the status sequence number of the PDU does not
match the expected status sequence number. For error recovery level zero,
this will cause the initiator to do a Session level logout. |
Number of SCSI Async events received | I | The number of asynchronous
events received by the initiator. |
Number of "target requests logout" Async
events received | I | The total number of times the target
sent an asynchronous event with the AsyncEvent set as "target requests
logout". |
Number of "target will drop connection"
Async events received | I | The total number of times the target
sent an asynchronous event with the AsyncEvent set as "target will
drop connection". |
Number of "target requests parameter
negotiation" Async events received | I | The total
number of times the target sent an asynchronous event with the AsyncEvent
set as "target request parameter negotiation". |
Number of "vendor specific" Async events
received | I | The total number of times the target
sent an asynchronous event with the AsyncEvent set as "vendor specific
Async event". |
Number of Reject PDUs due to Data Digest
errors | C | The total number of Reject PDUs
sent by the target that had the Reason set as "Data (payload) Digest
Error". |
Number of Reject PDUs due to SNACK rejects | I | The total number of Reject PDUs sent
by the target that had the reason set as "SNACK Reject". |
Number of Reject PDUs due to Protocol
Errors | D | The total number of Reject PDUs sent
by the target that had the Reason set as "Protocol Error ". |
Number of Reject PDUs due to excessive
Immediate Commands | T | The total number of Reject PDUs
sent by the target that had the Reason set as "Immediate Command
Reject". This typically happens if the target has too many outstanding immediate
commands. |
Number of Reject PDUs due to "Task in
progress" | T | The total number of Reject
PDUs sent by the target that had the Reason set as "Task in Progress". |
Number of Reject PDUs due to Invalid
Data ACK | D | The total number of Reject PDUs sent
by the target that had the Reason set as "Invalid Data ACK". |
Number of Reject PDUs due to Invalid
PDU field | D | The total number of Reject
PDUs sent by the target that had the Reason set as "Invalid PDU
Field". |
Number of Reject PDUs due to Lack of
Target Resources | T | The total number
of Reject PDUs sent by the target that had the Reason set as "Long Operations
Reject". A large value indicates that the target is frequently
running out of resources. |
Number of Reject PDUs due to Negotiation
Resets | I | The total number of Reject PDUs
sent by the target that had the Reason set as "Negotiation Reset". |
Number of Reject PDUs due to target Waiting
for Logout | I | The total number of Reject
PDUs sent by the target that had the Reason set as "Waiting for Logout". |
Number of Reject PDUs due to Miscellaneous
reasons | T | The total number of Reject PDUs sent
by the target for which the Reason code does not match any of the
currently defined codes. |
Time when statistics were last cleared | I | The time that the statistics
were last cleared. This provides an indication of the period of time
to which the statistics can be applied, and therefore can be used
for averaging the statistics. Because each system is different, separate
statistic rates can be determined on a per-system basis and used
to determine load changes. |