From: Rich Glazier (rglazier2002@yahoo.com)
Date: Mon Dec 08 2003 - 13:23:44 EST
Thanks to the three respondents. Everybody sort of
chipped in with different pieces of helpful
information, so I just included all 3 responses. Note
that some of Joe's correspondence is VMS centric.
-----
John Lanier wrote
Hello Rich,
I am speaking from a support perspective (I believe I
have worked with
you before; I believe Dennis Kettles is the TAM for
your Gold account? ),
but I am running webes 4.3. Be aware that both ca
and wsea are EV6-centric,
meaning that they will only report accurate info. for
the Alpha EV6 platform
(DS/ES/WF platforms). Also, dia and uerf may report
some info. on the EV6 platform but I
wouldn't trust the callouts too much; it's only
recommended to use dia/DECevent on EV5 and
lower systems.
"man wsea" and "man webes" has some good info. as
well, but if you are
used to DIA/DECevent syntax, then working with wsea is
the same as working with
ca in that inserting a "u" for "unix" (see ex.) will
work for doing various
queries.
EX:
====
wsea u -o summary
wsea u -i cam -o summary
wsea u -R|more
wsea u -R -f binary.errlog|more
etc..
I hope this helps somewhat.
Regards,
--John L.
Alan Rollow wrote
I use SEA a little, though I mostly write WEBES/SEA
rules for the Enterprise Virtual Array. As it
happens
I wrote our regression test for these rules to run on
a Tru64 UNIX system.
WEBES doesn't have rules for the system I use
(AlphaServer
1200), so its use for analysis of whatever happens
there
is limited. In your case, the DS20E does have WEBES
rules,
so using to monitor the system may notice problem
before
you do. The Storage rules handle disk events and a
variety
of Storage subsystem; backplane RAID controller, HSZ
family
subsystems, etc.
The browser interface is good for looking at problem
reports when it does find something and even the bit-
to-text section is more readable through the browser
than the CLI output. Previous versions had
performance
problems, but the ones we were made aware of have
been
addressed.
Since it has to be installed to be usable for our
regression
test, I have the local notification setup to send me
mail.
I haven't gotten a message from it yet, but since it
doesn't
have rules for the system, there isn't a lot for it
to do.
In the past, I've had trouble with what I suspect is
the
reader thread thinking there are event when there
aren't
and running continuously. I usually keep the
director
turned off for this reason, but that was using V4.2
and
earlier, so I should see if the problem has been
fixed
Joe Rudyk / Rich Glazier wrote (some information
applies to VMS only)
Hi Rich,
I am a little puzzled now. I thought you were using
VMS but it sounds
like you are using Webes on Tru64Unix. Is that so?
Well I can't say much
regarding Webes on Tru64Unix as I personally do not
manage those
systems, though we have some, but I am not the Unix
expert.
But maybe I missunderstood... so if you are using VMS
here are the
Answers to your questions..
- yes, the VMS ERRLOG.SYS is the equivalent to
/var/adm/binary.errlog
(I asked my Unix workmate)
the location typically is SYS$SYSROOT:[SYSERR] or
check the location
by issung the command
$SHOW LOGICAL SYS$ERRLOG
- the knowledge files are somewhere burried in the
directory tree of
Webes or CCAT. You can't use them by any other method
else than running the
CA (WSEA) comand. It is not like you could edit or
type them and see
anything usefull.
- AUTOCLUE is a tool that used to be on the VMS CD'S
but they
discontinued that. If you wanted it, I could provide
you with the VMS installation
kit, it is ca. 2MB in size.
If you are using VMS and only then!!! you find the
location of the
files under the directory CLUE$COLLECT, which is a
logical name. Issue the
command at DCL Prompt $SHOW LOGCAL CLUE* to see all
related logical names for
AUTOCLUE, like...
"CLUE$COLLECT" = "SYS$COMMON:[SYSERR]"
"CLUE$ENABLE_COLLECT" = "TRUE"
"CLUE$ERRLOG" = "SYS$SCRATCH:CLUE$ERRLOG.SYS"
"CLUE$FLAG" = "FFFF"
"CLUE$HISTORY" =
"SYS$COMMON:[SYSERR]CLUE$HISTORY.DAT"
"CLUE$MAX_BLOCKS" = "5000"
"CLUE$OUTPUT" = "SYS$MANAGER"
"CLUE$SDA" = "SYS$SHARE:CLUE$SDA"
"CLUSTER_DIR" = "FS8102:[SYSTEM]"
Typically the files that are created after a crash are
named like this
CLUE$nnnnnn_ddmmyy_hhmm.LIS
where nnnnnn is the nodename (usually the DECnet
nodename)
ddmmyy is the date in the format day,month,year in
two digits
hhmm is the time
- CCAT cann not be use manually, as there is no CLI
for it. However,
you can manually call the startup procedure by issuing
the command
$@SYS$STARTUP:CCAT$STARTUP.COM
- I mixed two things in my first mail, sorry about
that!!!!- CCAT does
not create the CLUE file!!! Autoclue creates the file
in CLUE$COLLECT see
further text. The CCAT startup procedure is called
every time after a
boot or crash and checks for reasons of a crash and
creates an automatic
call with HP, giving some hints and reasons for the
crash. You can still use
it for yourself by giving in your email address during
installation.
I hope this helps a little, and let me know if you
want the OpenVMS
AutoClue
Kits!
Cheers Joe
-----Ursprungliche Nachricht-----
Von: Rich Glazier [mailto:rglazier2002@yahoo.com]
Gesendet: Dienstag, 2. Dezember 2003 18:11
An: Rudyk, Josef
Betreff: Re: WEBES 4.3
Joe,
Thnaks a lot for all the feedback- it was very
informative. I have a few followup questions, if you
don't mind.
-I could not find a file named ERRLOG.SYS- is this the
same as my /var/adm/binary.errlog?
- Where are the knowledge files to translate the
errlog file, and do they ever need to be updated (or
are they automatically updated with new firmare or
something)?
- Where is CLUE$COLLECT an AUTOCLUE? So CCAT
automatically analyzes the crash files after a system
crash? Can I run it manually to look at vmzcore?
Thanks for all the help!
--- "Rudyk, Josef" <Josef.Rudyk@commerzbank.com>
wrote:
> Hi Rich,
>
> this Joe Rudyk from Commerzbank Frankfurt / Germany.
>
>
> Yes we are useing Webes. The version we use is 4.1-1
> and we use it on
> various systems like DS25, ES45 and GS160. I don't
> know how familiar you are
> with this tool..? So let me explain some more even
> if you know it already.
>
> 1) Do not use Webes on NON-GSx, -ESx and -DSx
> systems, like AS4100 or AS8200
> for example. I've tried the other week with Webes
> 4.3 on an AlpahServer 8200
> and it was using 200% CPU, i.e. 2 out of 4 CPUs. The
> info from HP was, that
> Webes is not meant to run on ASxxxx systems, for
> those, DECevent is still
> the one and only tool.
>
> 2) Webes is/can be helpfull even if you don't use
> the automatic call logging
> like we do. Webes contains knowledge files that
> translate the binary error
> messages from the ERRLOG.SYS into readable text
> format. So it gives you
> important information about problems and errors.
> DECevent has, as far as I
> know, not the capability to fully translate the
> errorlog from GS, ES and DS
> systems as HP changed the binary format for that HW.
>
> 3) CCAT creates a file in CLUE$COLLECT after a crash
> using AUTOCLUE. It
> summarizes the event, reading/analyzing the
> dumpfile.
>
> 4) WSEA is merely another name... it basically is
> still the same as CA
> before.
>
> 5) When using Webes, CA or WSEA, make sure, that you
> rename your ERRLOG.SYS
> file every midnight. If you don't, it'll grow over
> time. From my experience
> from anything between a few hundred blocks per day
> to a few thousend blocks
> depending on the number of events and how much
> hardware is involved, say...
> a few hundred disks? By renaming it every midnight,
> you keep it at a
> 'normal' size which makes analyzing it much quicker.
>
> OK... I hope, this has answered some of your
> questions.
>
> Regards ... Joe
__________________________________
Do you Yahoo!?
Free Pop-Up Blocker - Get it now
http://companion.yahoo.com/
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:46 EDT