ArcServe: Manutenzione del database di ArcServe 2000

Questo articolo spiega come usare le utilità di gestione del database di ArcServe 2000.

Riferimenti

• How to Maintain the ARCserve 2000 Database
• ArcServe 2000 Sample Script for ARCserve VLDB Maintenance
• Sample batch/CMD script for BrightStor ARCserve/Enterprise Backup VLDB maintenance

Pack del Database

• Assicurarsi che non ci sia nessun job attivo
• Assicurarsi che non ci sia apert ArcServe Manager
• Digitare i comandi:

%CONNPATH%dbdefrag -L arcserve;admin;secret astpsdat
%CONNPATH%keybuild -k -L arcserve;admin;secret astpsdat

Verificare che la dimensione del datase sia diminuita.

Messaggi di Errore tipici

• Failed to open database xxxxxxxx:error 10.

RESOLUTION: The database is being accessed by another process, preventingthe utility from locking the database. Make sure the database pruning job isnot running, that no jobs are merging data into the database, that the ARCservemanager is closed, and try again.

---

How to Maintain the VLDB Database

Please note that much of the information in thisarticle applies to ARCserveIT 6.61 VLDB as well. The new VLDB database that is used in ARCserve2000 is the latest generation of the ARCserve Raima database. It has a capacityof 360 million records (roughly 36 gigabytes of data) and also boasts improvedresponse and access time.

VLDB Databases

The VLDB is composed of 7 databases, the one thatmany administrators will be most concerned about is the astpsdat database. Thisholds most of the information that is displayed in the restore screen and willusually be the largest database of the VLDB. The VLDB databases are:
asjob
asmedia
asobject
asrhost
astape
astpdrv
astpsdat

Checking the Database Type (For ARCserveIT 6.61and 7.0 (2000) users only)

Before going on, check the internal database typebeing used on your server. The two internal ARCserveIT database types, the originalRaima database, and the new VLDB database, use very different maintenance utilities,so it is important to make sure that you know which type you are using. To check this, go into the ARCserveIT manager,and select the Database window. The database type will be indicated on the right,under the heading "Local Database Type". If the type is "ARCserveITVLDB" then you are using the new VLDB database. If the database type is"ARCserveIT" or "ARCserve", you are using the standard Raimadatabase, and should refer to Document 16112 instead.

Maintaining database size

There are two main factors that control databasesize in ARCserve. The first is the database pruning operation, and the secondis the pack function, which is executed with the dbdefrag utility.

How to remove records using Database Pruning

If the database has become larger than desired,it is possible to reduce the number of records in the database. This can bedone by either the Database Pruning Job or by manually deleting entries fromthe ARCserve Manager Database window. Manually deleting records via the ARCserveDatabase manager will free up space within the ARCserve Database, but it willnot reduce the database key file sizes.

Database Pruning Job:

The Database Pruning Job is a built-in ARCservejob that cannot be deleted from the Job Queue. It is enabled by default to remove,each night, any database records older than 30 days. The frequency of this jobcan be modified from the ARCserve Server Admin, which is located in the ARCserveManager's Quick Access screen (or the ARCserveIT Program Group for v6.61 and7.0 users).

File:TECH02857 as2k-1521-01.jpg

From within the ARCserve Server Admin, select theConfiguration icon and then the Database tab. There is a check box for enablingthe Pruning Job if it is not already. In addition, you can alter the numberof days a record resides in the database before it is removed. If desired, itcan be set so that only tape-related records are deleted. There is also a checkbox to run the Database Pruning Job "now". Note: The Database PruningJob will effect server utilization, possibly causing the server to slow down.If you a working with a production server, you may want to wait until afterhours to run the Pruning Job. Enable Database Pruning - default is on

Detailed information regarding the files and directories that were backed up or copied in a session are deleted when database pruningis performed. By default, this option is selected in order to free up spacein the database files. However, this option can be disabled to maintain thedetailed information for restoring purposes. Keep in mind, however, that notpruning your database can cause it to grow out of control.

File:TECH02857 as2k-1521-02.jpg

ARCserve Manager Database window:

From the ARCserve Manager, under the Database window,select either the Job or Tape View. Entries can be highlighted and deleted by clickingon the trashcan. Multiple entries can be selected at the same time by holdingdown the Ctrl key.

File:TECH02857 as2k-1521-03.jpg

Records will be deleted at the Job and Tape levelonly, and not at the session level. Deleting the records from the Database manageronly creates empty space within the database files. Database pruning is onlythe first step in reducing database size. This only marks entries as deleted(and therefore reusable), but they still take up space. The next step is torun the dbdefrag utility.

The VLDB Utilities for Database Maintenance:

Running Programs From The Command Prompt All VLDB utilities are command line utilities -and thus must be run from the command prompt. To run a VLDB utility from thecommand prompt:
Find out what directory ARCserve is installedin. If ARCserve is in the directory c:\arcserve, then to run a utility withinNT, go to START, programs, Command Prompt. It should come up with c:\>. Thecommand cd\arcserve (the directory for this example) will then take you to theARCserve directory. From here, any of the utilities can be run by copying thesyntax given in this document

Dbcheck (Perform this on a weekly basis)

Use this utility to run a check of a VLDB database's integrity. To run theVLDB dbcheck utility from the Command Prompt, in the ARCserve directory, usingthe syntax:
dbcheck -a -L arcserve;admin;secret DBNAME

where DBNAME is the name of the database that youwant to check the integrity of (i.e. asjob, asmedia, asobject, asrhost, astape,astpdrv, and astpsdat). As the check runs, information on the status of thedatabase should be displayed on the screen, ending with a final database summary. If a check does find a problem with a database,there are several options for repairing the corruption.

Keybuild (Perform this only if Dbcheck returnserror or if you do dbdefrag)

The keybuild utility repairs high-level keycorruption. The keybuild utility can be run from the Command prompt, in theARCserve directory, using the syntax:

keybuild -k -L arcserve;admin;secret DBNAME

where DBNAME is the name of the database that youwant to repair (i.e. asjob, asmedia, asobject, asrhost, astape, astpdrv, andastpsdat). As keybuild runs, the status of the operations should be displayedon the screen, ending with a final summary of the rebuilding process.

Dbfix (Perform this on a biweekly basis) =

Dbfix will check each entry in the database for corruption and repair anyerrors found in the database. This can be run from the Command Prompt, in theARCserve directory, using the syntax:

dbfix -a -L arcserve;admin;secret DBNAME

where DBNAME is the name of the database that youwant to repair (i.e. asjob, asmedia, asobject, asrhost, astape, astpdrv, andastpsdat). As dbfix runs, the status of the operations should be displayed onthe screen, ending with a final summary of the repair process. Dbdefrag (Run this utility only if you decreasedpruning time and you want to make sure that you have as much free space as thelargest database file) - The dbdefrag followed by keybuild utility can be usedto compress a database after running a prune job, and it can also repair corruptioncaused by broken links. Within the VLDB entries that are linked, ( i.e. A->B->C->...),if a link is corrupted or destroyed, the VLDB will no longer be able to accessdata.Running a dbdefrag may remove thesebroken links and thus allow the VLDB to access this data

* PLEASE NOTE* The dbdefrag utility should onlybe run after a prune job has run and removed records
(if there are no records marked as deleted, the defrag will have nothing todo). The VLDB defrag utility can be run from the CommandPrompt, in the ARCserve directory, using the syntax:

dbdefrag -L arcserve;admin;secret DBNAME

where DBNAME is the name of the database that youwant to defrag (i.e. asjob, asmedia, asobject, asrhost, astape, astpdrv, andastpsdat).

VLDB utility troubleshooting

Below are some common error messages that can bereceived when running the VLDB utilities and how to correct the problem. PROBLEM: Failed to log into server 'arcserve',username 'admin': error -2005.
RESOLUTION: The VLDB utility can notlog into the database. The most likely causes are that the ARCserve databaseengine needs to be started or the server name entered is incorrect (it shouldbe arcserve). PROBLEM: Failed to log into server 'arcserve',username 'admin': error -1002. Failed to log into server 'arcserve', username'admi': error -1001.
RESOLUTION: The VLDB utility can not log into the database because ofincorrect login information. Make sure when running a utility that the commandincludes: -L arcserve;admin;secret This enters in login information (arcserve,which is the database server name; admin, which is the user name; and secret,the default admin password). This security information can only be used whenrunning the VLDB maintenance utilities on the local ARCserve server (where theVLDB database is located) and only allow access to the VLDB utilities. PROBLEM: Failed to open database xxxxxxxx:error 10.
RESOLUTION: The database is being accessed by another process, preventingthe utility from locking the database. Make sure the database pruning job isnot running, that no jobs are merging data into the database, that the ARCservemanager is closed, and try again. PROBLEM: Failed to open database xxxxxxxx:error -40.
RESOLUTION: An invalid database name has been entered. Double check thatthe correct database name has been used when entering the command (i.e. asjob,asmedia, asobject, asrhost, astape, astpdrv, and astpsdat).

Debugging Database Engine

Modify the registry to add the following value:
Hkeylocal machine\computerassociates\arcserve \ base\database\-->>
edit\add
value=DebugTapeApi
data type=reg_dword
data=1

AND

Modify the registry to add the following value:
Hkeylocal machine\computerassociates\arcserve\base\database\>>
edit\add
value=debug
data type=reg_dword
data=1
Debugging info is added to arcserve.log.
What is Database Corruption?
Database corruption can be defined as the point at which ARCserve can no longerperform a Retrieve, Insert, Update or Delete of a record within a database file.It is possible that a database file, like any other computer file, may becomecorrupt. If this happens, an error message from ARCserve will be displayed toalert the user to the problem. Depending upon when the problem occurred, andwhat process ARCserve was performing at the time, one of several error messageswill be displayed.
If you look in the Activity log, do you see an asterisk (*) next to the JobIDnumber? If so, then this indicates that the ARCserve Database is not being writtento. If this is the case, then there may be database corruption. If you do notsee an asterisk, then see if there are any *.CAT files in the ARCSERVE\TEMPdirectory. If there are then there seems to be a problem Merging the CAT fileinto the database.
How Does Database Corruption Occur?
Database corruption may occur ifaccess to the database file is abruptly interrupted during a Retrieve, Insert,Update or Delete of a record within the file. What Are the Most Common Ways in Which aDatabase May Become Corrupted: Not maintaining VLDB database - Like anyother relational database VLDB database must be maintained regularly using VLDButilities (Dbcheck, keybuild, Dbfix etc).
Server Crash - If the ARCserve Server crasheswhile a database file is open and being processed, database records will notbe cached correctly.
Downing the Server - If the ARCserve Serveris downed while a database file is open and being processed, database recordswill not be cached correctly.
Host Server Runs Out of Disk Space - Ifthe ARCserve Server runs out of disk space while a database file is open andbeing processed.
CPU usage is peaked - If the server usageis peaked during a database file is open and being processed, database recordswill not be cached correctly.
Viruses Attacked the server - If the serverwas attacked by a virus while a database file is open and being processed, databaserecords will not be cached correctly.
Excessive Network/Server related activity- Could also be a reason for database corruption.
Reached the database limit - VLDB databasehas a capacity of 360 million records (roughly 36 gigabytes of data).
Pruning- Having very high pruningtime will cause the database to grow very rapidly and might reach its limit.We use approximately 80 bytes of hard drive space per file per backup in thedatabase.
Merge time - Setting the merge time to coincidewith backup time may cause database corruption
Merge time & Prune time - Setting pruningtime to coincide with merging may cause database corruption.
Velocis.ini settings - Having non-optimalsettings in the Velocis.ini file may increase chances of database corruption
Fix Corruption:
Try stopping the ARCserve Engines and then deletethe following files, found in the database directory.
The three files are: rdm.chk - checkpoint image file
raaaaaaa.chg - Change log file
rdm.chi - Change log indicator file These files are index files that will be recreatedthe next time ARCserve starts. Sometimes these files become corrupt and causethese types of problems.
If that does not fix the problem, then there maybe database corruption. Follow these steps to try and fix the corruption:
Edit the Velocis.ini file, it is found inthe ARCSERVE\DATABASE subdirectory.
Below is a portion of an average Velocis.ini, change the AsyncRead value asfollows: [SERVER]
ServerName=ARCSERVE
Archiving=1
AsyncRead=1 <--- Change the 1 to 0
AsyncTransactions=0
AsyncWrite=1
ChangeLogBufSize=262144
ChangeLogDevice=catdev
CheckPointDevice=catdev
CheckPointThreshold=40
MaxCachePages=22400 [default is 5600]
MemoryLimit=134217728 [default is 33554432] What this is doing is changing the way Raima processesthreads. Basically, the default way is to "queue" the threads to beprocessed at a later time. When this is done, this can cause a bottleneck andthe symptom is Failed to Login to db ec=1034. When you change the AsyncReadto 0, you are telling Raima to process its jobs immediately so that there isno queue. Sometimes the Database file RDS.EXE hangs up. This file should unloadwhen you stop the ARCserve Database Engine. If it doesn't you can reboot theserver or Use the NT3.51 resource Kit to Kill the RDS.exe. With ARCserve 6 you can Check and repair corrupteddatabase by executing the Dbmgr.exe in the ARCserve home directory. If a databaseis corrupt and will not repair you will be forced to Re-initialize the Databasefrom the ARCserve Server Admin. Click on the Database Engine Tab then clickon Operation and a menu option will drop down. You will see the Initialize databaseoption.

---