257 lines
10 KiB
Plaintext
257 lines
10 KiB
Plaintext
#
|
|
# $Id: hwrouter.help,v 1.12 1999/05/11 19:26:03 kenmcd Exp $
|
|
#
|
|
|
|
@ hinv.nrouter number of CrayLink routers in the system
|
|
The number of CrayLink routers in the system that have one or more
|
|
connected ports.
|
|
|
|
@ hinv.nrouterport number of connected CrayLink router ports in the system
|
|
The total number of connected ports on all CrayLink routers in the
|
|
system.
|
|
|
|
@ hinv.map.router path to a CrayLink router in the hardware graph
|
|
The path to a CrayLink router in the hardware graph filesystem.
|
|
|
|
There is one string-valued instance of this metric for each CrayLink
|
|
router configured in the system.
|
|
|
|
@ hw.router.portmask active port mask for each CrayLink router
|
|
Non-zero bits in the value of this metric indicate CrayLink router
|
|
ports that are connected in the system. Zero bits indicate the
|
|
corresponding router port is not connected.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
@ hw.router.rev_id router chip version id number for each CrayLink router
|
|
The CrayLink router chip version identification number.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
@ hw.router.send_util average send utilization for each CrayLink router
|
|
The send utilization averaged over all connected ports on a CrayLink
|
|
router.
|
|
|
|
A value of 100% indicates all connected ports on a particular CrayLink
|
|
router are sending at maximum capacity. This is independent of the
|
|
receive utilization (see hw.router.recv.total_util).
|
|
|
|
In practice, it is rare to observe sustained CrayLink router
|
|
utilization exceeding 30%. See the description for
|
|
hw.router.perport.send_util for details.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
@ hw.router.recv.total_util average receive utilization for each CrayLink router
|
|
The receive utilization averaged over all connected ports on a CrayLink
|
|
router.
|
|
|
|
A value of 100% indicates all connected ports on a particular CrayLink
|
|
router are receiving at maximum capacity. This is independent of the
|
|
send utilization (see hw.router.send_util).
|
|
|
|
In practice, it is rare to observe sustained CrayLink router
|
|
utilization exceeding 30%. See the description for
|
|
hw.router.perport.recv.total_util for details.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
@ hw.router.recv.bypass_util average receive bypass utilization for each CrayLink router
|
|
The average of hw.router.perport.recv.bypass_util for all connected
|
|
ports on each CrayLink router.
|
|
|
|
See hw.router.perport.recv.bypass_util for details.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
@ hw.router.recv.queued_util average receive queued utilization for each CrayLink router
|
|
The average of hw.router.perport.recv.queued_util for all connected
|
|
ports on each CrayLink router.
|
|
|
|
See hw.router.perport.recv.queued_util for details.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
@ hw.router.retry_errors total retry errors for each CrayLink router
|
|
The total number of retry errors on all connected ports for each
|
|
CrayLink router in the system. This counter is normally converted to a
|
|
rate (per-second) by client tools.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
Retry errors are not fatal, however persistent occurrence of these
|
|
errors may indicate a CrayLink interconnect problem. Refer to
|
|
hw.router.perport.retry_errors to identify the particular ports that
|
|
are contributing to the aggregated count for the router.
|
|
|
|
@ hw.router.sn_errors total sequence number errors for each CrayLink router
|
|
The total number of sequence number errors on all connected ports for
|
|
each CrayLink router in the system. This counter is normally converted
|
|
to a rate (per-second) by client tools.
|
|
|
|
On some early versions of the CrayLink router and hub this metric may
|
|
include some normal transactions, i.e. some non-sequence errors may be
|
|
incorrectly counted as sequence errors.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
Do not be alarmed by sequence number errors, these are expected
|
|
in normal operation and not fatal.
|
|
|
|
@ hw.router.cb_errors total checkbit errors for each CrayLink router
|
|
The total number of checkbit errors on all connected ports for each
|
|
CrayLink router in the system. This counter is normally converted to a
|
|
rate (per-second) by client tools.
|
|
|
|
There is one instance of this metric for each CrayLink router in the
|
|
system.
|
|
|
|
Checkbit errors are not fatal, however persistent occurrence of these
|
|
errors may indicate a CrayLink interconnect problem. Refer to
|
|
hw.router.perport.cb_errors to identify the particular ports that
|
|
are contributing to the aggregated count for the router.
|
|
|
|
@ hw.router.perport.send_util send utilization for each CrayLink router port
|
|
The utilization of the send bandwidth for each CrayLink router port,
|
|
computed from the statistically sampled averages reported by the
|
|
hardware.
|
|
|
|
A value of 100% indicates a port has reached its maximum send
|
|
capacity. CrayLink routers support independent send and receive
|
|
channels on each port and so this metric is independent of the
|
|
corresponding receive utilization (see hw.router.recv.total_util).
|
|
|
|
In practice, it is rare to observe sustained CrayLink router port
|
|
utilization exceeding 30%.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
@ hw.router.perport.recv.total_util receive utilization for each CrayLink router port
|
|
The utilization of the receive bandwidth for each CrayLink router port,
|
|
computed from the statistically sampled averages reported by the
|
|
hardware.
|
|
|
|
A value of 100% indicates a port has reached its maximum receive
|
|
capacity. CrayLink routers support independent send and receive
|
|
channels on each port and so this metric is independent of the
|
|
corresponding send utilization (see hw.router.send_util).
|
|
|
|
In practice, it is rare to observe sustained CrayLink router port
|
|
utilization exceeding 30%.
|
|
|
|
Values for this metric equal hw.router.perport.recv.bypass_util added
|
|
to hw.router.perport.recv.queued_util utilization for each CrayLink
|
|
router port.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
@ hw.router.perport.recv.bypass_util receive bypass utilization for each CrayLink router port
|
|
The utilization of the receive bandwidth for each CrayLink router port
|
|
for packets that could be processed without first being queued in the
|
|
DAMQ (Dynamically Allocated Memory Queue). Packets which bypass the
|
|
DAMQ queue incur lower transmission latencies.
|
|
|
|
Values for this metric added to the corresponding values for the
|
|
hw.router.perport.recv.queued_util metric equal the value of the
|
|
hw.router.perport.recv.total_util metric.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
@ hw.router.perport.recv.queued_util receive queued utilization for each CrayLink router port
|
|
The utilization of the receive bandwidth for each CrayLink router port
|
|
for packets that could not be processed without first being queued in
|
|
the DAMQ (Dynamically Allocated Memory Queue). Packets which do not
|
|
bypass the DAMQ incur higher transmission latencies.
|
|
|
|
Values for this metric added to the corresponding values for the
|
|
hw.router.perport.recv.bypass_util equal hw.router.perport.recv.total_util.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
@ hw.router.perport.retry_errors retry errors for each CrayLink router port
|
|
The number of retry errors for each connected port on each CrayLink
|
|
router in the system. This counter is normally expressed as a rate
|
|
per-second by client tools.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
Retry errors are not fatal, however persistent occurrence of these
|
|
errors may indicate a CrayLink interconnect problem.
|
|
|
|
@ hw.router.perport.sn_errors sequence number errors for each CrayLink router port
|
|
The number of sequence number errors for each connected port on each
|
|
CrayLink router in the system. This counter is normally converted to a
|
|
rate (per-second) by client tools.
|
|
|
|
On some early versions of the CrayLink router and hub this metric may
|
|
include some normal transactions, i.e. some non-sequence errors may be
|
|
incorrectly counted as sequence errors.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
Do not be alarmed by sequence number errors, these are expected
|
|
in normal operation and not fatal.
|
|
|
|
@ hw.router.perport.cb_errors checkbit errors for each CrayLink router port
|
|
The number of checkbit errors for each connected port on each CrayLink
|
|
router in the system. This counter is normally converted to a rate
|
|
(per-second) by client tools.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
Checkbit errors are not fatal, however persistent occurrence of these
|
|
errors may indicate a CrayLink interconnect problem.
|
|
|
|
@ hw.router.perport.excess_errors excessive errors flag for each CrayLink router port
|
|
The number of error "types", i.e. retry or checkbit, that exceed an
|
|
effective rate of 500 per minute over the last polling period.
|
|
|
|
A non-zero value (i.e. 1 or 2) indicates excessive errors on the port.
|
|
|
|
There is one instance of this metric for each connected port on each
|
|
CrayLink router in the system.
|
|
|
|
@ hinv.map.routerport path to a CrayLink router port in hardware graph
|
|
The path to a CrayLink router port in the hardware graph filesystem.
|
|
|
|
There is one string-valued instance of this metric for each connected
|
|
port on each CrayLink router in the system.
|
|
|
|
@ hinv.interconnect interconnection endpoint in hardware graph filesystem for each CrayLink router port
|
|
There is one string valued instance of this metric for each connected
|
|
port on each CrayLink router in the system. The metric's value is the
|
|
path to the destination node or destination CrayLink router port in the
|
|
hardware graph filesystem.
|
|
|
|
The values for this metric, in conjunction with the external instance
|
|
identifiers (i.e. CrayLink router ports), are sufficient to determine
|
|
the connection topology of an Origin series system.
|
|
|
|
Considering nodes and CrayLink routers to be vertices in the system
|
|
topology and CrayLink router ports to be arcs, the external instance
|
|
identifiers provide the name of the source of each link (i.e. a
|
|
specific port on a CrayLink router) and the metric instance values
|
|
provide the name of the destination (i.e. another CrayLink router or an
|
|
Origin series node).
|
|
|
|
@ hw.router.type Router type
|
|
The type of Origin router. Possible values are:
|
|
|
|
0 - normal router with at most two nodes, or a star router with up to 4 nodes
|
|
1 - metarouter which has only router connections for larger configurations
|