----------------------------------------------------------------------
[Contents]
Attached you will find an [Overview] which defines the savepnpc
problem we were trying to resolve.
We also included a brief explanation on [What is savepnpc] with links
to a savepnpc man pages.
The obligatory [Thank You] section which acknowledges the people who
responded to or who showed interest in our [Q] query.
A [Summary] section written in a narrative email style.
A description of a [Common savepnpc hang-up] and how to get unhung.
Plus we including two (2) bonus sections on configuring NSR to
automatically shutdown and restart Oracle databases. We got it to
work in the [How to for Digital UNIX] and [How to for Windows NT]
sections.
Plus you will find our original [Q] AND [Q continued] sections at the
bottom of this [Summary].
---------------------------------------------------------------------
[Overview]
Networker savepnpc will execute client-scripts at the beginning of a
backup group and then execute client-scripts at the completion of a
backup group. We were having problems making NSR's savepnpc facility
work us. Specifically we could not get savepnpc to shutdown or
restart our Oracle8 databases. This problem was resolved.
Our thanks go out to those who helped us or showed interest in our [Q]
query for assistance. We feel the problem we experienced was on the
cutting edge because not many people from the mailing list responded
with suggestions, but lucky for us, we were able to resolve this
problem through our support agreement with Compaq. See the file on
ticket #C990329-2340.
The real resolution for this bug lies either with Legato 5.2 and/or
Oracle 8.0.4 because it is really their bug, however you can apply a
patch locally to fix this problem at your site it you choose to. Jump
down to the [Summary] section to find out about this patch.
Oracle users take note! My DBAs assure me if your not doing one of
the following three scenarios, NSR is not giving you a good backup of
your database. To successfully backup your Oracle database with NSR,
you must be doing one of the following:
1) Invest in the Oracle Business Suite so you can do "HOT ON THE
FLY" backups and other good things.
2) Use the "savepnpc" feature to automate database shutdowns and
restarts before/after NSR performs its "COLD" off-line backup of
your database.
3) If you choose not to use the Business Suite, or "COLD" automation
through savepnpc you must manually shutdown your database instance
before your backup process begins and then manually restart that
instance upon completion of the backup. Suppose your backup group
contains (10) ten clients and each client has a database, you must
manually shut down each client's database instance before the
backup group runs. If you aren't off-line-ing these databases you
aren't getting a good backup of them. That's how I understand it
and that's the main assumption for this [Q] and [Summary] review.
To resolve our savepnpc problems, I think we logged three separate
calls on the subject because we had several problems to resolve,
however once you learn the tips, tricks, and trade-secrets of
savepnpc, getting it to work is really quite easy.
----------------------------------------------------------------------
[What is savepnpc]
savepnpc is a feature of Legato's NSR beginning with release 5.2.
This feature provides Legato customers a mechanism for executing
scripts prior to backing up a client group and for executing scripts
at the completion of the backup for that client group. You create
your backup groups, giving them names, and deciding which clients
belong to that group. It is very easy to add or remove one or more
clients from a given backup group using the nwadmin GUI.
You need savepnpc to shut down processes that are running because we
believe NSR skips over resources attached to running processes, and if
you shut a process down you'll probably want to restart it after your
backup completes. Savepnpc does this for you. :)
(********* savepnpc in a nutshell **********)
http://www.adcomp.ch/cluhelp/appb37.htm#50217
----------------------------------------------------------------------
[Thank You]
John Speno speno_at_isc.upenn.edu
Partin.Kevin KPartin_at_hou.mdc.com
Russ Fish Russ_Fish_at_idx.com
Harish Arora arisharora_at_hotmail.com
Gene Zurik zurik_at_alf.dec.com
Lannie Loh loh_at_decatl.alf.dec.com
Sakellaris Alexander asakelaris_at_cosmote.gr
Mike Gaunnac MGaunnac_at_DWD.State.IN.US
----------------------------------------------------------------------
[Summary] [Q] NSR savepnpc - what am I doing wrong 02/26/1999
---
Kevin,
If you find reasonable answer to this problem then please forward it
to me also. I am also looking for this functionality. TIA!
-Harish Arora 02/26/1999
---
---
Harish,
I will but I am not having any luck with this thing yet, and its
really starting to bug me. Savepnpc looks so easy to use too! I don't
know what I'm doing wrong either. Cause I followed the directions. :(
-Kevin
---
---
Kevin,
Are these commands working by hand? If yes, then use full path
instead of changing directory by cd command ie sh /sbin/init.d/oracle
stop
-Alexander Sakellaris 02/26/1999 ---
---
Alex,
You are probably on to something with this "full path" suggestion of
yours. I am just not ready to hear what you are saying. Sorry, I am
just kind of dense and stupid, maybe next time I will listen to you
sooner. :)
-Kevin 02/26/1999
---
---
I give up. I'm posting another query to the mailing list and logging
another service call to Compaq on this.
-Kevin 03/29/1999
---
---
Compaq Resolution #C990329-2340 03/29/1999
Hi Kevin,
I took a short lunch and looked at your [Q]continued web page.
This is what I think has happened...
Since it couldn't find "su", "data", and "touch" command, your pre
command failed. When it failed, it wouldn't create
a /nsr/tmp/<groupname>.tmp file. When the second saveset started,
savepnpc though the pre command hasn't been run yet, so it ran again.
Then it failed again... This repeated for all savesets. I think
this is why you see savepnpc tried to run your script multiple times.
Please try to use full path for those commands to see if it works.
Please let me know the result.
Thanks,
-Lannie
---
---
Lannie,
So what you are saving is, all the commands in the script
/sbin/int.d/oracle need to be fully qualified in order to make
Networker's savepnpc facility function correctly.
Is this due to some deficiency in the way Legato's coded their
savepnpc facilities or could it be some environmental setting in our
.dtprofile or .profile as it relating to my PATH variables? I don't
think the paths /usr/opt/networker/bin or /usr/var/nsr/res are set in
my PATH statements for the NSR administrators.
Reference: NSR: savepnpc what am I doing wrong now?
http://www.ornl.gov/its/archives/mailing-lists/alpha-osf-managers/1999
/03/msg00611.html
-Kevin
---
---
Hi Kevin,
If you don't use full path, you need to define your PATH environment
variable correctly in your script.
Savepnpc runs your script as root, but I don't think it should inherit
PATH from root login since it didn't run root's login file, .profile
or whatever it's using, and it shouldn't run. This is just like you
run a script thru cron. The default PATH for root is very minimum.
If you print the value of the PATH in your script, you'll see what you
get.
Thanks,
-Lannie
---
---
Lannie,
Your suggestion appears to be working. I am not sure who wrote the
/sbin/init.d/oracle script. I assume it was the vendor any way one of
our DBA's, Ramesh, made the changes you suggested and it appears to
have resolved our savepnpc problems.
Lannie, you may close out the problem ticket. Thank you for working
our problem. We really appreciate it.
-Kevin
---
---
Hi Kevin,
I am glad that the problem was solved, and thanks for letting me know.
Have a nice day!
-Lannie
---
---
We believe the root of the problem lies somewhere between the Legato's
5.2 and Oracle's 8.0.4 release. We hope somebody will read this
summary and address the problem in future releases. Until they do
this, you need to address it locally. Edit your /sbin/init.d/oracle
script and fully qualify the paths to the commands: "date", "touch",
and "su" to path locations appropriate for your shop..
-Kevin
---
----------------------------------------------------------------------
[common savepnpc hangups]
On 02/24/1999 we noticed savepnpc quit working for us. We could tell
it was hung by looking at date and time stamp information in the
/nsr/logs/messages and /nsr/logs/savepnpc.log files. They stopped
accruing new messages. If this happens to you it means your savepnpc
pre and post command scripts are not getting executed. Your hung up.
These hang-ups can occur to one or more clients of a given group while
other groups work correctly without a glitch. If this happens you
need to check locally at each client for the presence of
/nsr/tmp/your-NSRgroup-name.tmp and remove those files.
---
Dear Compaq,
Savepnpc quit working on 02/24/1999. I have the command savepnpc
entered in the client's Backup Command attribute, I believe this use
to work but recently it quit working. We are just now ramping
Networker into a production environment after a couple of months
experimentation. I believe savepnpc quit working for us on
02/24/1999. Can you help?
- Kevin
---
---
Kevin,
Here is a suggestion that Lannie had. Check for the presence of the
following files.
/nsr/tmp/group.tmp where group is your group name
If it exists, remove it and delete the group.res files from /nsr/res
and let nsr create the default files. Then
wait and see what happens.
- Gene Zurik problem resolution #C990305-1359 ---
---
Gene,
Yes that fixes my problem. I guess I hosed NSR somehow, and those
temporary files got stuck. I checked all my clients for stray
/nsr/tmp/*.tmp files and removed all tmp files I could find. Savepnpc
is working again.
-Kevin
---
----------------------------------------------------------------------
[How to for Digital Unix]
1) Using the nwadmin GUI create a group and add your clients to that
group. This is a fairly easy thing to do, you just play with it.
You can't hurt anything playing with this.
2) Using the nwadmin GUI, for each client you want to use savpnc with
go to {client} {client setup} pulldown menu and highlight the
desired client. Then type the following word in that client's
backup command field >> savepnpc <<. Don't type in these >> <<
characters though.
3) What was the name of the group you created in step 1? lets call
it your-group-name.
On each client in that group with a database create this file
/nsr/res/your-group-name.res
Then edit this file /nsr/res/your-group-name.res and type in the
following commands.
---
type: savepnpc;
precmd: "/usr/bin/sh /sbin/init.d/oracle stop", "/bin/sleep 60";
pstcmd: "/usr/bin/sh /sbin/init.d/oracle start", "/bin/sleep 60";
timeout: "12:00pm";
---
That's it end of step 3, don't keyin the --- characters.
4) There is a bug somewhere between Legato 5.2 and Oracle 8.0.4 you
can correct it though. Apply this local patch. Edit the
following file /sbin/init.d/oracle. Fully qualify the paths to
the "su", "date", and "touch" commands. Fix-qualify all
references to these commands.
Ours commands use to look like this: Now they look like this:
su /usr/bin/su
date /usr/bin/date
touch /usr/bin/touch
That's it end of step 4.
5) You are ready to go, don't experiment during business hours
because you will bring down your database. Using the nwadmin GUI
touch on the group button, highlight the your-group-name group
and press the start button. Watch out your database is going
down, its going to get backed up and restarted too!
----------------------------------------------------------------------
[How to for Windows NT]
---
From Mike Gaunnac
3/2/99 11:40:15 AM (rev. 3/19/99 11:24:41 AM)
If you want to use the NET START and NET STOP commands of Windows NT
to control your Oracle Database:
1. Set service OracleService<SID> to start manually
2. Set service OracleStart<SID> to start automatically
3. Set registry entry HKLM/SOFTWARE/ORACLE/ORA_SHUTDOWN = TRUE
4. Shutdown and restart the instance
5. In command mode :
NET STOP OracleTNSListener80
NET STOP OracleStart<SID>
NET STOP OracleService<SID>
NET START OracleStart<SID>
NET START OracleTNSListener80
These commands may be used to bracket a cold-backup session.
6. For Networker set up the file C:\WIN32APP\NSR\RES\<group_name>.RES
as follows:
type: savepnpc;
precmd: "NET STOP OracleTNSListener80","NET STOP
OracleStart<SID>","NET STOP OracleService<SID>";
pstcmd: "NET START OracleStart<SID>","NET START OracleTNSListener80";
timeout: "12:00pm";
See "Oracle8 Enterprise Edition Getting Started" Chapter 10 if you
need further instructions.
- Mike Gaunnac
---
---
Thanks Mike, I can see your script is working. Thanks a million, I
appreciate your help.
- Kevin
---
----------------------------------------------------------------------
[Q] NSR: savepnpc - Continued, what am I doing wrong now? 03/29/1999
Below you will find our group resource file. The goal is to shutdown
the Oracle8 databases before the backup job begins. Below you will
find the savepnpc group resource file that we experiment with.
/nsr/res/Alpha123-backup.res
type: savepnpc;
precmd: "/usr/bin/sh /sbin/init.d/oracle stop", "/bin/sleep 60";
pstcmd: "/usr/bin/sh /sbin/init.d/oracle start", "/bin/sleep 60";
timeout: "12:00pm";
Diagnostics produced after running this group would indicate the
pre-command is executing once for each save set instead of "AS
ADVERTISED" once for each client in the backup group.
There are (3) clients in the backup group. He are sample diagnostics
of what we are getting.
--- Unsuccessful Save Sets ---
* alpha1.dwd.state.in.us:/ 1 retry attempted
* alpha1.dwd.state.in.us:/ past first iffi set *
alpha1.dwd.state.in.us:/ LOG config
* alpha1.dwd.state.in.us:/ /sbin/init.d/oracle: touch: not found *
alpha1.dwd.state.in.us:/ Test of stop begins here
* alpha1.dwd.state.in.us:/ /sbin/init.d/oracle: date: not found *
alpha1.dwd.state.in.us:/ stopping V2 listner
* alpha1.dwd.state.in.us:/ stopping Oracle database
* alpha1.dwd.state.in.us:/ /sbin/init.d/oracle: su: not found
* alpha1.dwd.state.in.us:/usr /sbin/init.d/oracle: touch: not found *
alpha1.dwd.state.in.us:/usr Test of stop begins here
* alpha1.dwd.state.in.us:/usr /sbin/init.d/oracle: date: not found *
alpha1.dwd.state.in.us:/usr stopping V2 listner
* alpha1.dwd.state.in.us:/usr stopping Oracle database
* alpha1.dwd.state.in.us:/usr /sbin/init.d/oracle: su: not found
* alpha1.dwd.state.in.us:/a1u00 /sbin/init.d/oracle: touch: not found
* alpha1.dwd.state.in.us:/a1u00 Test of stop begins here
* alpha1.dwd.state.in.us:/a1u00 /sbin/init.d/oracle: date: not found
* alpha1.dwd.state.in.us:/a1u00 stopping V2 listner
* alpha1.dwd.state.in.us:/a1u00 stopping Oracle database
* alpha1.dwd.state.in.us:/a1u00 /sbin/init.d/oracle: su: not found .
.
.
etc. etc. etc. for each save set of all (3) clients
---
Additional Info follows:
All thre clients run NSR 5.2, under Digitial UNIX 4.0E with no patch
kits applied. Plus each client has two network interface cards
configured with the host names for these containing either mixed case
letters and/or hyphens. I.E. alpha1, alpha1-nsr, alpha2, alpha2-nsr,
alpha3, alpha3-nsr.
Also we are not using the Business Suite module for Oracle8 even
though the licenses packs were purchased, enabled and authorized
through NSR. We don't because our DBA's figure why do hot on the fly
backups when we have the time at night for cold offline backups.
---
- Thanks
Kevin Criss
----------------------------------------------------------------------
[Q] NSR: savepnpc what am I doing wrong? 02/26/1999
http://www.ornl.gov/its/archives/mailing-lists/alpha-osf-managers/1999
/02/msg00615.html
Received on Tue Apr 06 1999 - 22:44:07 NZST