ADSM-L

Re: LOADDB and RESTORE DB fails

1998-05-28 09:31:29
Subject: Re: LOADDB and RESTORE DB fails
From: Dwight Cook <decook AT AMOCO DOT COM>
Date: Thu, 28 May 1998 08:31:29 -0500
     Well, my tests have been during disaster recovery drills and on an MVS
     system... We have to many data base files to list on the :
     //SERVER EXEC PGM=DSMSERV,DYNAMNBR=300,
     //    PARM='/INSTALL ................'
     The maximum length of a continued EXEC JCL statement was exceeded so I
     would have to run this with just one log & one dbfile, then start adsm
     and perform "define dbvol" & "define logvol" commands for the other 8
     dbvols and 6 logvols (sigh).  The first time I did this I didn't
     extend things and the restoredb failed (don't remember exactly what it
     said) then I seem to remember I did complete extends and it still
     didn't like something so I just listed in my disaster recovery
     document to make things exactly as they were before.  To ensure I have
     this information there is one AIX box whose main job is to pull in
     such information from all 13 of our adsm environments about every 6
     hours.

     I will say we drive our adsm environments to the extreme, so I have
     situations to deal with that won't be found in 99% of the adsm
     environments out there...  (my favorite from a unix client is "why
     does it take so long to run an incremental on a file system with 1
     million files in it?"  my answer is generally "do an 'ls -l' and call
     me back when it finishes!" )

     later,
           Dwight


______________________________ Reply Separator _________________________________
Subject: Re: LOADDB and RESTORE DB fails
Author:  PromptSoftware (PromptSoftware AT T-ONLINE DOT DE) at unix,mime
Date:    5/27/98 12:44 PM


Yes, we moved the physical server to another place.
When it wasn't starting we allocated the log and db2 volumes the same size,
the same names and the same directories as before.
When we do a restoredb the server reports a reduction of the size of the
databse to 900MB (it was originally 1GB and 100MB still to extend).

But what do you mean with 'issue extends to make it as before' ? I thougt
after issueing
DSMSERV INSTALL 1 E:\ADSMDBASE\LOG01.DSM 2 E:\ADSMDBASE\DB01.DSM 500
E:\ADSMDBASE\DB02.DSM 500
the database and logs are full extended.

Thanks

Gerhard





     You moved the physical server ? Kept it on the same machine and
     attempted to restart it ?

     Oh, one thing I noticed in my testing you must extend the DB & LOG to
     the same size they were when you saved the DB... I've only tested
     restoredb from backupdb, not dumpdb's.
     I would allocate the DB & LOG files with the same characteristics as
     before but didn't perform extends and things would crap out on me.  I
     then re-did all the allocations and issued extends to make it as
     before and things loaded just fine.

     Try what I mentioned above with a restoredb and you should be OK.

     OH, you do have like a "Q DB F=D" and a "Q LOG F=D" don't you ?

     later
          Dwight


______________________________ Reply Separator ____________________________
_____
Subject: LOADDB and RESTORE DB fails
Author:  PromptSoftware (PromptSoftware AT T-ONLINE DOT DE) at unix,mime
Date:    5/27/98 9:36 AM


Environment: ADSM Server V2R1M12 on OS/2

Problem: After halting the server, shutting down the machine and putting it
 to another location  dsmserv failed to start with a
failore ANR9999D dballoc.c... Zero bit count (sorry, I don't have the exact
 message available).

Next we tried a DSMSERV DUMPDB DEVCL=MANTAPE and it worked, but brought
some errors at the end about something it can't read.

After making a new database with DSMSERV INSTALL 1 E:\ADSMDBASE\LOG01.DSM 2
 E:\ADSMDBASE\DB01.DSM 500 E:\ADSMDBASE\DB02.DSM 500
we tried to load the database dump with DSMSERV LOADDB DEVCL=MANTAPE
VOL=DSM054 and we get the following messages

 ANR4039I LOADDB: Loaded 274000 database entries (cumulative).
 ANR4039I LOADDB: Loaded 275000 database entries (cumulative).
 ANR4039I LOADDB: Loaded 276000 database entries (cumulative).
 TABLE ERROR 777 - Please get the following information to ADSM
development:
 Col = 0
 type1 = 2 (x2)
 t ype2 = 3 (x3)

 Key 1 (ascii) ->(3C0C)(00000000)<-
 Key 2 (ascii) ->(CF000000)(00000000)<-

 Key 1 (ebcdic) ->(3C0C)(00000000)<-
 Key 2 (ebcdic) ->(CF000000)(00000000)<-

 Key 1 (hex) ->(3C0C)(00000000)<-
 Key 2 (hex) ->(CF000000)(00000000)<-

 ANR9999D pkthread.c(741): Run-time assertion failed: "type1 == TBCOL_NULL
||
 type2 == TBCOL_NULL", Thread 0, File tbcol.c, Line 761.
 ANR7836S Server initialization terminated.
 ANR7837S Internal error TBCOL777 detected.


After several tries with new databases on different drives we decided to
restore an old database backup.
Creating a new database and starting DSMSERV RESTORE DB TODATE=TODAY we
get:

 ANR4639I Restored 45888 of 60530 database pages.
 ANR4639I Restored 49744 of 60530 database pages.
 ANR4639I Restored 53600 of 60530 database pages.
 ANR4639I Restored 57456 of 60530 database pages.
 ANR4640I Restored 60530 pages from backup series 6 operation 0.
 ANR0306I Recovery log volume mount in progress.
 ANR9999D dballoc.c(823): Zero bit count mismatch for SMP page addr 9216;
Zero
 Bits = 601, HeaderZeroBits = 0.
 ANR0361E Database initialization failed: error initializing database page
 allocator.
 ANR9999D icrest.c(1257): Unable to initialize database.
 ANR1364I Input volume DSM013 closed.
 ANR2032E RESTORE DB: Command failed - internal server error detected.

Then we upgraded the server to V2R1M14 --> the same.

Does anybody know, what's going on, how to get out of this situation ?
Why to make database backups, when I can't use them later on ?


Thanks

Gerhard
<Prev in Thread] Current Thread [Next in Thread>