Informix OnLine Database

Backup of Informix OnLine Database

Whole-Database Method

Incremental Method

Changing Size of Informix OnLine Database

Additional Databases

Informix OnLine Database

This chapter describes how to manage the Informix OnLine Database.

Backup of Informix OnLine Database

Two methods are available for backup and retrieval of your Cisco WAN Manager (CWM) Informix OnLine database. They are the Whole-database method, and the Incremental method.

The Whole-database method backs up the entire Informix OnLine database. This method is used for databases requiring only one or two 150-megabyte tapes (the standard Sun tape). You also use this method when you want to increase the size of the database.

The Incremental method allows root, first-level incremental, and second-level incremental backups. The Incremental method is recommended when you want to keep constant incremental backups, as well as when you have databases greater than 200 megabytes. This method does not work for changing the size of the database.

Whole-Database Method

When the Informix OnLine database is either full or in danger of being full at a future date, you can create a one-time backup of the database, increase the database's size, and restore the data.

The Whole-Database method is also more convenient than the Incremental method when you have a database of 200 or fewer megabytes:

Step 1 Shut down CWM by selecting Option 2 (Stop Core) from the CWM Main Menu.

Step 2 When changing the size of your Informix OnLine database, edit the onconfig file found in the /usr/users/informix92/etc/onconfig directory, changing the ROOTSIZE parameter to the new value.

Step 3 Either log out and log back in as user informix, or rlogin to the workstation where CWM is installed.

Step 4 Invoke the onunload command. For example:

host% ./onunload -t /dev/rst0 -s 144000 stratacom

This indicates tape device rst0 and a tape size of 144 megabytes (which is appropriate for a standard 150-megabyte Sun tape). You are prompted to insert one or more additional tapes, as necessary.

Step 5 When the Informix OnLine database has been backed up, and you do not intend to restore it immediately, you are done. To change the Informix OnLine database size, proceed to Step 6. To restore the Informix OnLine database, proceed to Step 7.

Step 6 To change the size of the database, log in as user root, change directory to the /usr/users/svplus directory, and invoke the INSTALL command at the command line prompt. This re-installs the database. Typically, three files fail a checksum test: start_SV+, svmain, and stat.err. You can ignore these checksum failures, as these files do change subtly during regular use. Type Yes to the prompt asking whether the database should be erased. The following error message (which can be ignored) is displayed: "create_db: failure drop database! -371". Proceed to Step 9.

Step 7 To restore the Informix OnLine database, log in as user informix. Log in again as superuser, then change directory to the /usr/users/svplus directory.

Step 8 You must delete the existing database. (Normally, you are restoring the old database because corruption or a disk error has occurred; therefore, the current data is not viable.) Enter the following commands:
host% ./kill_db
host% ./init_db
host% ./ create_db

Step 9 Log out as superuser. You are once again logged in as user informix. Change directory to the /usr/users/informix72 directory, then invoke the dbaccess command. The dbaccess program, like other Informix OnLine maintenance programs, provides menu items at the top of the screen. You select a menu option by moving to it using the left and right arrow keys, and then pressing <RETURN>, or by typing the first letter of the option. Select Database (either by pressing D or by pressing the right arrow key to Database and then hitting <RETURN>). Next, select Drop. One database is listed. It is called stratacom. Press <RETURN> to select it. Confirm the database drop by typing Y. You have now dropped the database. Type E twice (for Exit) and the UNIX prompt is displayed.

Step 10 You can now reload the database by using the onload command. For example:
host% ./onload -t /dev/rst0 -s 144000 stratacom

This indicates tape device /dev/rst0 and a tape size of 144 megabytes.

Step 11 You can now log back in as user svplus and run the CWM product with the restored database.

Incremental Method

For larger databases, or databases for which you would prefer Incremental backups, the archive process used by the Informix OnLine onmonitor program is recommended. Running this program provides menus similar to those provided by the dbaccess program.

For archiving, the CWM program can continue to run. You must first open a terminal window, for example a C-shell window.

Step 1 Remotely login (rlogin) as user informix to the workstation where CWM is installed.

Step 2 Invoke the onmonitor command:
host% ./onmonitor

Step 3 Select A for Archive, then T for Tape-Parameters. In the tape size field, specify the number, in kilobytes, of the size of your tape. (144000 is correct for the standard 150-megabyte Sun tape.) Press <ESC> then C to select Create.

Step 4 You are prompted to insert a tape. You are then asked whether this is a Level-0, 1, or 2 archive. Level-0 archives collect the entire database. Level-1 archives capture everything since the last Level-0 archive. Level-2 archives capture everything since the last Level-1 archive. Note, the first time you do an incremental backup, you must select Level-0.

Step 5 A highlighted line is displayed indicating the size of the Informix OnLine database space. This is always the total size configured for the database, not the amount of data in the Informix OnLine database. Press <control-B> to continue. When more than one tape is required, you are prompted to load the subsequent tapes. At the end, answer N to the prompt about backing up logs. When back at the Archive menu, press E (Exit) twice and log out as user informix. Your Informix OnLine database is now backed up. Instructions for restoring the archive begin with Step 6.

Step 6 To restore an archive, CWM cannot be running. Also note, the entire database is replaced with the archived data, so any non-archived data will not remain after the restoration. Stop CWM by specifying Option 2 (Stop Core) at the CWM Main Menu.

Step 7 Remotely login (rlogin) as user informix to the workstation where CWM is installed.

Step 8 Invoke the onmonitor command:
host% ./onmonitor

Step 9 The Informix OnLine database must be off-line for a restoration. Select M (for Mode), then T (for Take-Offline), and then type E (for Exit) to return to the Main menu. The Informix OnLine database is now in off-line mode.

Step 10 Select A (for Archive) and then R (for Restore) from the Main menu. When you are restoring a Level-2 archive, you also must restore Levels 0 and 1 corresponding to that archive. When you are restoring a Level-1 archive, you also must restore Level-0. When you are restoring a Level-0 archive, you only need to restore the Level-0 tape(s).

Step 11 Insert the Level-0 tape first, then press <control-B> when ready. Answer Y to the "Do you have another Level to restore" question, if you wish to restore higher-level archives as well. At the Archive menu, enter E (for Exit) to return to the Main menu.

The Informix OnLine database is now in quiescent mode. Enter M (for Mode), then O (for On-Line), and then E (for Exit) twice to exit the onmonitor program. Log out as user informix. You can now restart CWM with the restored data.

Changing Size of Informix OnLine Database

Changing the size of the Informix OnLine database eradicates the data stored in the database. Therefore, before changing the size of the database, you should save it.

Even when you don't want to change the size of the database, it is a good idea to back it up periodically.

The /usr/users/informix72/bin/tbmonitor program provides an archiving facility allowing you to back up the database. Informix can be in on-line or quiescent mode to execute an archive. When you are archiving while the system is on-line, be sure to choose an off-peak period.

Step 1 Log in as user informix.

Step 2 Change directory to the /usr/users/informix72/bin directory:
host% cd /usr/users/informix72/bin

Step 3 Run the tbmonitor program:
host% ./tbmonitor

Step 4 Select the Main menu's Archive option.

Step 5 Select the Archive menu's Create option.

You are prompted to mount the tape.

Step 6 Insert a write-enabled tape into the tape drive. (Make sure the tape drive is on line.)

Informix OnLine then prompts you for the level of archiving you desire.

Level-0 copies all dbspaces to tape.
Level-1 or Level-2 records changes since the last archive.

: When a Level-0 archive has already been done, you can choose Level-1, Level-2, or Level-0. When a Level-0 archive has not already been done, you must enter 0. Wait while the archiving proceeds. (Tapes rewind automatically when full.) The percentage of completion of the archival process is displayed on the screen and is updated at one-minute intervals.
: When the first tape is full, the following messages are displayed:
: Please label this tape as number 1 in the archive sequence.
Please mount tape and press <RETURN> to continue.

Step 7 Label the completed tape and place a new write-enabled tape into the tape drive.

Step 8 Repeat steps 5 and 6 until the backup is complete.

When the archive process is complete, the message "Program over" is displayed.

Additional Databases

You can load a non-default database by selecting the Currently db name option from the Cisco WAN Manager (CWM) Main Menu during CWM start-up.

To create a new database, as user root, type:
host# create_db -D <databasename>

Note Subsequently created databases share the same disk space as the default database. When the addition of other databases results in insufficient space, you may need to add another CWM workstation to the network.

After creating a new database by executing the create_db command, if CWM is restarted before it is synchronized with the network, it will result in a database inconsistency. In this case, perform create_db before restarting CWM to avoid database inconsistencies.

Note Do not use the SV+CreateDb tool to create a new CWM database.

The logical_connection table count for a node in the CWM database may not match the master lcon counts in that node. This happens when the database contains FastPAD connections, or when the database has slave lcons with no connections.

When a Gateway node is upgraded or rebuilt, before CWM has synced up with the network, CWM must be cold-started (create CWM database and restart CWM). This is required to ensure CWM database consistency.

The CWM database and network may differ by value 1 for parameters PCR(0+1), SCR(0+1), MCR, ICR of some ATM-FR connections. This happens because of the difference (rounding effect) in conversion algorithms used by CWM and the Switch to convert values in cells/sec units to 100 bit/sec units.

CWM CM GUI has text fields for the parameters PCR(0+1), SCR(0+1), MCR, and ICR (all in the unit of cell/sec) on the ATM side of the ATM - FR ABR.FS provision window.
The front-end application (CM GUI) passes these values to the back-end process to setup the SNMP request PDU for the Switch Agent. Following are the MIB variables and the associated value from the GUI:

: atmEndptPCR corresponds to PCR(0+1)
: atmEndptSCR corresponds to SCR(0+1)
: atmEndptMCR corresponds to MCR
: atmEndptQIR corresponds to ICR

After the SNMP set request to the switch is done, the values are deposited in the user_connection table. Following are the database table column names:

: l_pir (100 bit/sec) corresponds to PCR(0+1)
: l_cir (100 bit/sec) corresponds to SCR(0+1)
: l_mir (100 bit/sec) corresponds to MCR
: l_qir (100 bit/sec) corresponds to ICR

As the unit for the four columns is 100 bits/sec, conversion from cells/sec to 100 bits/sec is done.
The conversion algorithm results in 100 bps = (384 * cps) + 99) / 100 as shown in the following table:

cell/second 100 bit/second

50

192

5760

22119

3259

12515

2834

10883

cell/second	100 bit/second
50	192
5760	22119
3259	12515
2834	10883

Note The converted values for 5760, 3259, and 2834 are rounded up to the nearest 100 bits/sec.

The conversion factor used by the switch (in this case BPX 9.2) is different from that used by CWM. Therefore, after the ConnObj Robust Message update, the user_connection table has different values for those four columns. See the following table with example values:

.

Cell/second CWM
100 bit/second Switch
100 bit/second

5760

22119

22118

3259

12515

12514

2834

10883

10882

Cell/second	CWM 100 bit/second	Switch 100 bit/second
5760	22119	22118
3259	12515	12514
2834	10883	10882

The CWM database may intermittently have incomplete connections for various reasons. When a multi-segment PVC is added, and a node receives a connection add request, but doesn't respond to CWM, then CWM backs out all of the segments from the other nodes which have responded to the add request, and deletes those segments from the CWM DB. The node that did not respond ends up having that segment. Because of this, any further attempt to add that connection, after the node becomes reachable, results in a "Connection already exists" error in CWM. However, this connection segment may or may not appear in the CWM database, depending upon which node was not responsive. Following are two example scenarios:

Figure C-1: Two Segment Connection

Figure C-2: Three Segment Connection

: In Figure C-1 and Figure C-2, F1 and F2 are feeder shelves, and S1 and S2 are routing nodes. If S1 becomes unreachable after receiving an add request from CWM, but before sending the response back, Middle segment will be added on S1, between S1 and S2. However, this connection's segment details will not be in the CWM database. Once S1 becomes reachable, configuration changes will be sent to CWM and the CWM database will be updated with the middle segment details. You can delete the incomplete connection at this time, using CWM.
: However, if F1 (or F2) is an MGX 8220 shelf and it becomes unreachable after receiving an add connection segment request, CWM does not get updated with the end segment details, even if the node becomes reachable. The database update happens whenever the next periodic Node Resync by CWM happens with that shelf.

Similar DB inconsistency can happen in the case of deleting a connection operation. In this case, while a multi-segment PVC is being deleted, if a node receives a connection delete request, but doesn't respond to CWM, CWM doesn't delete this segment information from the CWM database. This results in a dangling connection segment in the CWM database, which doesn't exist on the node. However, this connection segment will be deleted from the CWM database eventually. The time taken for this database sync-up, once the node becomes reachable, depends on the type of node.

: In the above two examples, if S1 becomes unreachable after receiving a delete request from CWM, but before sending the response back, Middle segment will be deleted on S1, between S1 and S2. However, this connection's segment details will not be deleted from the CWM database. Once S1 becomes reachable, configuration changes will be sent to CWM and the CWM database will be updated accordingly.
: However, if F1 (or F2) is an MGX 8220 shelf and it becomes unreachable after receiving a delete connection segment request, CWM does not delete the end segment details from the CWM database, even if the node becomes reachable. The database update happens whenever the next periodic Node Resync by CWM happens with that shelf.

Table of Contents