HP3000-L Archives

July 1998, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jeff Vance <[log in to unmask]>
Reply To:
Jeff Vance <[log in to unmask]>
Date:
Thu, 9 Jul 1998 23:48:50 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (214 lines)
Hi all,

Thanks to you who offered your opinions and recommendations regarding our
proposed changes to the DSTAT command.  Below we have summarized your input
and have offered a revised proposal.  There are still a couple of questions
remaining, sprinkled throughout this posting.

[ There are 3 questions, did you see them? }


Objectives:
-----------
  (1) Enhance DSTAT to handle up to 5 digit LDEV numbers and model numbers
  greater than 7 characters.  Long term we expect model numbers to be up to
  28 characters long.

  (2) Enhance DSTAT to show HAFO (High Availability Failover) information
  for dual ported disks like the Nikes.


Feedback Summary:
-----------------
* It's OK to increase LDEV number and model number fields but don't
  add any additional info to the default DSTAT output. Output will
  continue to be in upper case.  Specifically, it is ok to shift the
  default output 1 or 2 characters to the right to accomodate 4 and 5
  digit LDEV numbers. It is ok to increase the model field from 7 bytes
  to a fixed size of 12 bytes.  Note: today, DSTAT shifts one character
  right for 4 digit LDEVs.  Model numbers greater than 12 characters can
  be silently truncated.  Thus, the longest DSTAT line could look like:

today:
   LDEV-TYPE    STATUS      VOLUME (VOLUME SET - GEN)
  ----------   --------    --------------------------
  4-C3010M1     MEMBER      MEMBER4         (MPEXL_SYSTEM_VOLUME_SET-0)
 13-C3010M1     MEMBER      MEMBER13        (MPEXL_SYSTEM_VOLUME_SET-0)
131-C2490AM     MEMBER      MEMBER131       (MPEXL_SYSTEM_VOLUME_SET-0)
1234-C23456     CP*MEMBER-MDMEMBER35        (MPEXL_SYSTEM_VOLUME_SET-0)

Proposal:

new Choice 1:
   LDEV-TYPE       STATUS       VOLUME (VOLUME SET - GEN)
  --------------  ------------ --------------------------------------------
  4-C3010M1        MEMBER       MEMBER4         (MPEXL_SYSTEM_VOLUME_SET-0)
 13-C3010M1        MEMBER       MEMBER13        (MPEXL_SYSTEM_VOLUME_SET-0)
131-C2490AM        MEMBER       MEMBER131       (MPEXL_SYSTEM_VOLUME_SET-0)
1234-C23456        CP*MEMBER-MD MEMBER35        (MPEXL_SYSTEM_VOLUME_SET-0)
12345-C23456789012 CP*MEMBER-MD MEMBER35        (MPEXL_SYSTEM_VOLUME_SET-0)


new Choice 2:
   LDEV-TYPE       STATUS       VOLUME (VOLUME SET - GEN)
  --------------  ------------ --------------------------------------------
    4-C3010M1      MEMBER       MEMBER4         (MPEXL_SYSTEM_VOLUME_SET-0)
   13-C3010M1      MEMBER       MEMBER13        (MPEXL_SYSTEM_VOLUME_SET-0)
  131-C2490AM      MEMBER       MEMBER131       (MPEXL_SYSTEM_VOLUME_SET-0)
 1234-C23456       CP*MEMBER-MD MEMBER35        (MPEXL_SYSTEM_VOLUME_SET-0)
12345-C23456789012 CP*MEMBER-MD MEMBER35        (MPEXL_SYSTEM_VOLUME_SET-0)

The difference between choices 1 and 2 are:
o choice 1 keeps the LDEV column aligned as it is today for all rows with
  LDEVs three or less digits long.  However, 4 digit LDEVs break this rule
  today and 5 digit LDEVs will break it again tomorrow.

o choice 2 says bite the bullet and shift everything over to accomodate 5
  digit LDEVs thus impacting the column location for all LDEVs not just those
  that are 4 or more digits.  Small shops with low LDEVs could be affected
  by choice 2 and not by choice 1; however if we agree to increase the model
  field to 12 bytes then we have potentially impacted everyone anyways.

>>> QUESTION <<<
Do you prefer Choice 1 as the new default for DSTAT, or choice 2?

VOTE:
 ___ Choice 1
 ___ Choice 2
 ___ Don't care
 ___ other
 ___ Don't change current output at all


* It is OK to abbreviate the STATUS field in the default case if we
  don't have sufficient space to display bigger model names.  However, today
  the status field can contain 12 characters of data, as shown above.  We
  don't see an oppertunity to shorten the width of this field.

* Don't stagger the output of a single disk across multiple lines in
  the default case.

* The default output will just truncate the model number if it's too
  big and we will not display any character to indicate truncation.

* With the new DETAIL option, display HAFO info, disk size (in GB/MB ???)
  and %free space of every volume (not volume set) also.

>>> QUESTION <<<
What should the units of measure be for the new disk capacity field?

VOTE:
 ___ megabytes (2^20)
 ___ gigabytes (2^30)
 ___ other


* Use lower case for the output in the new DETAIL format.

* Align the data for the output with DETAIL option so that everything
  starts on the same column.

* Any field that doesn't have significance for a particular volume
  should be left blank. Don't display NONE or NA etc.


Below are all of the fields and their max size (width) for the new
DSTAT FORMAT=detail output:
------------------------

 LDEV:           nnnnn (5)
 MODEL:          xxxxxxxxxxxxxxxxxxxxxxxx (24)
 VOLNAME:        xxxxxxxxxxxxxxxx (16)
 VOLTYPE:        xxxxxxxxxxxx (12)
 VOLSET-GEN:     xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx-n (34)
 DISK SIZE:      xxxxGB (4)   or  xxxxxxxMB (7)
 PRIMARY PATH:   xxxxxxxxx (9)
 PRIMARY STATUS: xxxxxxxx (8)
 ALT PATH:       xxxxxxxxx (9)
 ALT STATUS:     xxxxxxxx (8)
 EVENT TIME:     hh:mm:ss (8)

 Important HAFO related fields
 -----------------------------

  PRIMARY PATH - path string for ldev

  ALTERNATE PATH - alternate path (DAM path) to which I/O will be
  rerouted for failover.

  PRIMARY STATUS - status or operational readiness of primary path.
  Status could be: array failure, failover failed, ready, test failover,
  unknown failure, unknown status

  ALTERNATE STATUS - status for alternate path. Status could be:array
  failure, failover failed, ready, unknown failure, unknown status, not
  validated

  EVENT TIME - formatted date and time of failover event. Events will
  be loged in the sysdiag file with additional time information.

 The ALT PATH, ALT STATUS and EVENT line will be displayed only for
 HAFO devices. These fields will be left blank for non-HAFO disks.

>>> QUESTION <<<<
or, should we just not display the ALT PATH, ALT STATUS and EVENT rows
if the device is not a HAFO device???

VOTE:
 ___ Always display the above three rows even if they don't apply to a
     particular device (there will be no data after the label)
 ___ Don't display the rows/labels if they don't apply to a particular device
 ___ Don't care
 ___ other


 Here's what it would look like for several devices -- some HAFO,
 others not:

 :dstat all;format=detail

 LDEV:           31
 MODEL:          c2490am-nike
 VOLNAME:        member1
 VOLTYPE:        master
 VOLSET-GEN:     mpexl_system_volume_set-0
 DISK SIZE (gb): 28.7
 PRIMARY PATH:   8.0.0
 PRIMARY STATUS: ready
 ALT PATH:       16.0.0
 ALT STATUS:     ready
 EVENT TIME:

 LDEV:           32
 MODEL:          c2390am
 VOLNAME:        member2
 VOLTYPE:        member
 VOLSET-GEN:     mpexl_system_volume_set-0
 DISK SIZE (mb): 28700
 PRIMARY PATH:   8.0.1
 PRIMARY STATUS: ready
 ALT PATH:
 ALT STATUS:
 EVENT TIME:

 LDEV:           31
 MODEL:          c2490am-nike
 VOLNAME:        member1
 VOLTYPE:        master
 VOLSET-GEN:     mpexl_system_volume_set-0
 DISK SIZE (gb): 28.7
 PRIMARY PATH:   9.0.0
 PRIMARY STATUS: failover
 ALT PATH:       17.0.0
 ALT STATUS:     ready
 EVENT TIME:     09:09:09

------------------------------------------------------------

Thanks and regards,

Narayan, CSY
Jeff Vance, CSY

--

ATOM RSS1 RSS2