On 04/ 8/10 03:48 PM, Garrett D'Amore wrote:
I'd still like a record of the kstats in this case, albeit with Project
Private (or Uncommitted) binding.

Why bother for Project Private interfaces? Because in this case the
kstats are quite visible to end users (as are all kstats) via kstat(1M),
and so I think its useful to record the fact that these are explicitly
*not* supported. It will also help potential future developers who might
need the information in the stats, although it will also underscore to
them that any use will require a contract.

So while the list isn't *required*, it *is* desired (by me at least).

I'm struggling to see the desire for this case to provide that
documentation.  This case is documenting a change in smbstat output.
The fact that the kstat API is used internally is an existing
internal implementation detail, and ARC'ing details of the
internal implementation between smbsrv and smbstat for the purpose
of having a specification for something whose use is unsupported
seems strange.

If another project wants to consume the kstats, it would seem better
to document them with the contract rather than live in a cycle of
having to repeatedly supersede this case with new cases on every
trivial change between smbsrv and smbstat, even when smbstat output
is unafffected.  That seems like unnecessary ARC noise.

Alan

On 04/ 8/10 02:11 PM, Jose Borrego wrote:

-----Original Message-----
From: Alan Wright [mailto:alan.wri...@oracle.com]
Sent: Thursday, April 08, 2010 12:38 PM
To: Garrett D'Amore
Cc: psarc-...@sun.com; smb-eng...@oracle.com
Subject: Re: SMB/CIFS Statistics [PSARC/2010/120 FastTrack timeout
04/14/2010]

On 04/ 7/10 10:59 PM, Garrett D'Amore wrote:
> On 04/ 7/10 10:47 PM, Jordan Brown wrote:
>> Garrett D'Amore wrote:
>>>
>>> What's missing is an interface table. I think though that based on
>>> the statement that its not to be parsed ("for humans only"), the
>>> output is not-an-interface.
>>
>> Yes. The stability is the same as the existing command:
>>
>> |_____________________________|_____________________________|
>> | Interface Stability | Uncommitted |
>> |_____________________________|_____________________________|
>> | Utility Output Format | Not-an-Interface |
>> |_____________________________|_____________________________|
>>
>
> You haven't mentioned anything about underlying kstats. Is this how
> these stats are implemented? Should that be covered under an ARC
case as
> well (perhaps this one)?

smbstat has always created its output from undocumented kstats and,
as with the interface stability, changing that wasn't part of this
project. The purpose here is solely to expand and improve on the
command output. The underlying kstats have not been considered for
anything other than internal smbstat consumption. Publishing them,
particularly if they may be used in scripts, without taking the
change of scope into consideration and reviewing the design from
that perspective is not a good idea.

I'd prefer to take this as an RFE rather than delay/withdraw this
case while we go back and review the project with a different scope.

This is a good point. Please indicate if everybody is okay with this.

--

>>> On 04/ 7/10 05:51 PM, Jordan Brown wrote:
>>>> Template Version: @(#)sac_nextcase 1.70 03/30/10 SMI
>>>> This information is Copyright (c) 2010, Oracle and/or its
>>>> affiliates. All rights reserved.
>>>> 1. Introduction
>>>> 1.1. Project/Component Working Name:
>>>> SMB/CIFS Statistics
>>>> 1.2. Name of Document Author/Supplier:
>>>> Author: Jose Borrego
>>>> 1.3 Date of This Document:
>>>> 07 April, 2010
>>>>
>>>> 1.4. Name of Major Document Customer(s)/Consumer(s):
>>>> PSARC
>>>> CIFS team
>>>>
>>>> 1.5. Email Aliases:
>>>> 1.5.1. Responsible Manager: barry.greenb...@sun.com
>>>> 1.5.2. Responsible Engineer: jose.borr...@sun.com
>>>> 1.5.3. Marketing Manager:
>>>> 1.5.4. Interest List: cifs-t...@sun.com
>>>>
>>>> 4. Technical Description:
>>>>
>>>> Executive Summary:
>>>>
>>>> This case improves the statistics the SMB/CIFS server maintains
as
>>>> well as
>>>> the output of smbstat.
>>>>
>>>> Problem:
>>>>
>>>> Currently the only statistics the SMB/CIFS server maintains are
>>>> counters
>>>> keeping track of how many requests of each type were received,
the
>>>> number
>>>> of open files, the number of sessions (users logged in) and the
>>>> number of
>>>> connections (trees connected). Those statistics are not enough to
>>>> describe
>>>> the load being applied as well as how the server is responding to
it.
>>>>
>>>> Solution:
>>>>
>>>> In order to get a better picture of the state of the SMB/CIFS
server
>>>> the
>>>> following information can be maintained and provided through
smbstat:
>>>>
>>>> - Counters:
>>>> * nbt:
>>>> Number of SMB NetBIOS-over-TCP (NBT) sessions.
>>>>
>>>> * tcp:
>>>> Number of SMB TCP sessions.
>>>>
>>>> * users:
>>>> Number of users logged in.
>>>>
>>>> * trees:
>>>> Number of trees connected.
>>>>
>>>> * files:
>>>> Number of open files.
>>>>
>>>> * pipes:
>>>> Number of open pipes.
>>>>
>>>> - Throughput:
>>>> * rbytes/s:
>>>> Number of bytes received per second.
>>>>
>>>> * tbytes/s:
>>>> Number of bytes transmitted per second.
>>>>
>>>> * reqs/s:
>>>> Number of requests handled per second.
>>>>
>>>> * reads/s:
>>>> Number of read requests per second. This would be an aggregation
of
>>>> the following requests:SMB_COM_READ, SMB_COM_LOCK_AND_READ,
>>>> SMB_COM_READ_RAW and SMB_COM_READ_ANDX.
>>>>
>>>> * writes/s:
>>>> Number of write requests per second. This would ba an aggregation
of
>>>> the following requests: SMB_COM_WRITE, SMB_COM_WRITE_AND_UNLOCK,
>>>> SMB_COM_WRITE_RAW and SMB_COM_WRITE_AND_CLOSE.
>>>>
>>>> - Utilization:
>>>> * wcnt:
>>>> Average number of requests received but waiting for an SMB worker
>>>> thread to execute them.
>>>>
>>>> * rcnt:
>>>> Average number of requests being simultaneously executed by an
SMB
>>>> worker thread.
>>>>
>>>> * wtime:
>>>> Average time a request waits before an SMB worker thread starts
>>>> executing it.
>>>>
>>>> * rtime:
>>>> Average execution time of a request.
>>>>
>>>> * w%:
>>>> Percentage of the time during which at least one request was
>>>> waiting.
>>>>
>>>> * r%:
>>>> Percentage of the time during which at least one request was
being
>>>> executed.
>>>>
>>>> * u%:
>>>> Percentage of utilization of the SMB/CIFS server. This number is
>>>> defined as: rcnt / (Max Worker Threads).
>>>>
>>>> * sat:
>>>> Flag indicating if the server saturated in the past. Saturation
is
>>>> defined as: u% == 100%.
>>>>
>>>> * usr%:
>>>> Percentage of the time the processor(s) spent in user space.
>>>>
>>>> * sys%:
>>>> Percentage of the time the processor(s) spent in kernel space.
>>>>
>>>> * idle%:
>>>> Percentage of the time the processor(s) was(were) idle.
>>>>
>>>> - Requests:
>>>> For each type of request:
>>>> * %:
>>>> Percentage of that type of request.
>>>>
>>>> * rbytes/s:
>>>> Number of bytes received per second.
>>>>
>>>> * tbytes/s:
>>>> Number of bytes received per second.
>>>>
>>>> * req/s:
>>>> Number of requests handled per second.
>>>>
>>>> * rt-mean:
>>>> Average response time in seconds.
>>>>
>>>> * rt-stddev:
>>>> Standard deviation of the response time.
>>>>
>>>> To access this information the CLI smbstat must be modified and
>>>> allow the
>>>> following options:
>>>>
>>>> -r Display the statistics of the requests.
>>>>
>>>> -a Display the statistics of all the types of request (the 256 of
>>>> them) whether they are valid or not.
>>>>
>>>> -n Display in alphabetic order.
>>>>
>>>> -z Display the statistics of the requests actually received.
>>>>
>>>> -t Display the throughput of the SMB/CIFS server.
>>>>
>>>> -u Display the utilization of the SMB/CIFS server.
>>>>
>>>> -c Display the counters.
>>>>
>>>> A number can also be provided indicating the interval of time
between
>>>> refreshes. The first statistics displayed cover the time between
when
>>>> the server was started and the moment the statistics were
requested.
>>>> Subsequent refreshes cover the last time interval only.
>>>>
>>>> It should be noted that the output is all new replacing the old
one and
>>>> is for humans only.
>>>>
>>>> Examples of smbstat output:
>>>>
>>>> 'smbstat -ctu'
>>>>
>>>> nbt tcp users trees files pipes
>>>> 0 1 1 2 20 0
>>>>
>>>> rbytes/s tbytes/s reqs/s reads/s writes/s
>>>> 1.036e+02 1.298e+00 0.000e+00 0.000e+00 0.000e+00
>>>>
>>>> wcnt rcnt wtime rtime w% r% u% sat usr% sys% idle%
>>>> 4.317e-01 7.410e+00 2.461e-05 4.224e-04 31 100 0 no 0 76 24
>>>>
>>>> 'smbstat -r'
>>>>
>>>> code % rbytes/s tbytes/s req/s rt-mean rt-stddev
>>>> SmbCreateDirectory 00 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbDeleteDirectory 01 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbOpen 02 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbCreate 03 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbClose 04 0 0.000e+00 0.000e+00 0.000e+00 2.496e-04 8.003e-06
>>>> SmbFlush 05 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbDelete 06 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbRename 07 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbQueryInformation 08 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbSetInformation 09 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbRead 0A 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbWrite 0B 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbLockByteRange 0C 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbUnlockByteRange 0D 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbCreateTemporary 0E 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbCreateNew 0F 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbCheckDirectory 10 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbProcessExit 11 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbSeek 12 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbLockAndRead 13 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbWriteAndUnlock 14 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbReadRaw 1A 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbWriteRaw 1D 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbSetInformation2 22 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbQueryInformation2 23 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbLockingX 24 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbTransaction 25 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbTransactionSecondary 26 0 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> 0.000e+00
>>>> SmbIoctl 27 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbEcho 2B 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbWriteAndClose 2C 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbOpenX 2D 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbReadX 2E 0 0.000e+00 0.000e+00 0.000e+00 3.550e-04 1.856e-05
>>>> SmbWriteX 2F 100 4.607e+02 1.675e+00 0.000e+00 3.893e-04 2.691e-
07
>>>> SmbCloseAndTreeDisconnect 31 0 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00 0.000e+00
>>>> SmbTransaction2 32 0 0.000e+00 0.000e+00 0.000e+00 5.203e-04
8.612e-06
>>>> SmbTransaction2Secondary 33 0 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00 0.000e+00
>>>> SmbFindClose2 34 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbTreeConnect 70 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbTreeDisconnect 71 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbNegotiate 72 0 0.000e+00 0.000e+00 0.000e+00 4.625e-05
0.000e+00
>>>> SmbSessionSetupX 73 0 0.000e+00 0.000e+00 0.000e+00 1.145e-01
0.000e+00
>>>> SmbLogoffX 74 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbTreeConnectX 75 0 0.000e+00 0.000e+00 0.000e+00 5.851e-04
2.436e-05
>>>> SmbQueryInformationDisk 80 0 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> 0.000e+00
>>>> SmbSearch 81 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbFind 82 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> SmbFindUnique 83 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbFindClose 84 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbNtTransact A0 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbNtTransactSecondary A1 0 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> 0.000e+00
>>>> SmbNtCreateX A2 0 0.000e+00 0.000e+00 0.000e+00 5.154e-02 2.063e-
03
>>>> SmbNtCancel A4 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbNtRename A5 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbOpenPrintFile C0 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>> SmbWritePrintFile C1 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbClosePrintFile C2 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
>>>> 0.000e+00
>>>> SmbGetPrintQueue C3 0 0.000e+00 0.000e+00 0.000e+00 0.000e+00
0.000e+00
>>>>
>>>> 'smbstat -rz'
>>>>
>>>> code % rbytes/s tbytes/s req/s rt-mean rt-stddev
>>>> SmbClose 04 0 0.000e+00 0.000e+00 0.000e+00 2.445e-04 5.906e-06
>>>> SmbReadX 2E 4 0.000e+00 2.772e+01 0.000e+00 1.556e-03 7.878e-06
>>>> SmbWriteX 2F 96 7.162e+02 2.604e+00 0.000e+00 5.365e-04 3.913e-07
>>>> SmbTransaction2 32 0 0.000e+00 0.000e+00 0.000e+00 5.203e-04
8.612e-06
>>>> SmbNegotiate 72 0 0.000e+00 0.000e+00 0.000e+00 4.625e-05
0.000e+00
>>>> SmbSessionSetupX 73 0 0.000e+00 0.000e+00 0.000e+00 1.145e-01
0.000e+00
>>>> SmbTreeConnectX 75 0 0.000e+00 0.000e+00 0.000e+00 5.851e-04
2.436e-05
>>>> SmbNtCreateX A2 0 0.000e+00 0.000e+00 0.000e+00 7.250e-02 2.304e-
03
>>>>
>>>> 6. Resources and Schedule
>>>> 6.4. Steering Committee requested information
>>>> 6.4.1. Consolidation C-team Name:
>>>> ON
>>>> 6.5. ARC review type: FastTrack
>>>> 6.6. ARC Exposure: open



_______________________________________________
opensolaris-arc mailing list
opensolaris-arc@opensolaris.org

Reply via email to