Purpose
Displays virtual shared disk device
driver statistics of a node.
Description
The statvsd command
displays virtual shared disk statistics of a node. For example, on
a busy server an increasing number of "requests queued waiting for
a buddy buffer" is normal and does not necessarily imply a problem.
Of more value is the "average buddy buffer wait_queue size" which
is the number of requests queued for a buddy buffer when the statvsd command
was issued. See the "Examples" section for the meaning of output lines.
Security
You must be in the AIX® bin group to
run this command.
Exit Status
- 0
- Indicates the successful completion of the command.
- nonzero
- Indicates that an error occurred.
Restrictions
You must issue this command
from a node that is online in the peer domain. To bring a peer domain
online, use the startrpdomain command. To bring a particular
node online in an existing peer domain, use the startrpnode command.
For more information on creating and administering an RSCT peer domain,
refer to the RSCT: Administration Guide.
Standard Output
Current RVSD subsystem run level.
Examples
The following examples display
virtual shared disk device driver statistics.
- The header line indicates the version and release of the code.
For example:
VSD driver (vsdd): IP/SMP Version:4 Release:1
- The level of virtual shared disk parallelism defaults to 9 and
is the buf_cnt parameter on the uphysio call that the device driver
makes in the kernel. For example:
9 vsd parallelism
- The maximum IP message size in bytes. For example:
61440 vsd max IP message size
- The number of requests that had to wait for a request block. For
example:
61440 vsd max IP message size
- The number of requests that had to wait for a pbuf (a buffer used
for the actual physical I/O request submitted to the disk). For example:
0 requests queued waiting for a pbuf
- The number of requests that had to wait for a buddy buffer. A
buffer that is used on a server to temporarily store date for I/O
operations originating at a client node. For example:
2689 requests queued waiting for a buddy buffer
- The number of requests queued for a buddy buffer when the statvsd command
was issued. For example:
0 average buddy buffer wait_queue size
- The number of requests that a server has rejected, typically because
of an out-of-range sequence number or an internal problem. For example:
4 rejected requests
- The number of responses that a client has rejected. Typically
because a response arrived after a retry was already sent to the server.
For example:
0 rejected responses
- The number of requests that were placed on the rework queue. For
example:
0 requests rework
- The number of read requests that were not on a 64 byte boundary.
For example:
0 64 byte unaligned reads
- The number of requests that got a DMA shortage. This condition
would require the I/O operation to be executed in nonzero copy mode.
For example:
0 DMA space shortage
- The number of requests that have timed out. The current timeout
period is approximately 15 minutes. For example:
0 timeouts
- There are a fixed number of retries. The retries counters display
the number of requests that have been retried for that particular
"retry bucket." Numbers appearing further to the right represent requests
that have required more retries. When a request exhausts its number
of retries, it gets recorded as a timeout. For example:
retries: 0 0 0 0 0 0 0 0 0
0 total retries
- Sequence numbers are internally used by the device driver. These
numbers are managed by the device driver and the Recoverable virtual
shared disk subsystem. For example:
Non-zero Sequence Numbers
node# expected outgoing outcase? Incarnation:0
11 125092 0 |
11 Nodes Up with zero sequence numbers: 1 3 5 7 9 11 12 13 14 15 16
Location
/opt/rsct/vsd/bin/statvsd
Related Information
Commands: ctlvsd, vsdnode
Refer
to RSCT: Managing Shared Disks for information on tuning
virtual shared disk performance.