Stability of the System

From Multics Wiki
Jump to: navigation, search

While the DPS8M development team has been running Multics on the simulator for several months now, and those of us with long-running systems have enjoyed long periods of up-time, you should take care to not store anything of value on your Multics system unless you also have that data backed up somewhere else.

Warning: We have experienced simulator crashes where the contents of the Multics storage system is damaged. Salvaging the storage system allows Multics to be rebooted, but any data written in the bootload prior to the crash may be lost. Please see the following sections for recommendations to ensure the safety and integrity of your Multics storage system.

Run with DIRW Configuration Setting

By default, a cold boot of Multics (and by extension a QuickStart image) does not have the DIRW parameter enabled. This parameter causes "dirty" (written to) directory pages to be periodically flushed from memory to disk. We will probably enable this in the next version of the QuickStart image, but we strongly recommend you enable this parameter. To enable this parameter, boot your Multics system to BCE or shut down a running Multics to BCE. Use the "config" command to add a PARM card to the config deck. Here is an example of how to shut down your running Multics system and add this card:

M-> logout * * *

Ready
M-> 
CONSOLE: RELEASED

 1210  as   LOGOUT             IO.SysDaemon.z dmn cord   0: 0 $0.06 (looc)
 1210  as   LOGOUT             Backup.SysDaemon.z dmn bk   0: 1 $0.08 (looc)
 1210  as   LOGOUT             IO.SysDaemon.z dmn prta   0: 0 $0.06 (looc)
 1210  as   LOGOUT             Utility.SysDaemon.z dmn ut   3: 9 $14.35 (looc)
 1210  as   LOGOUT             Volume_Dumper.Daemon.z dmn vinc   0: 1 $0.08 (looc)
M-> shut

 1210  as   act_ctl_: shutdown, 9 0.21 0.13 0.00 0.34 80:37:47 $19601.79
1210.9  shutdown complete
bce (boot) 1210.9: M-> config

M-> /parm/

Search failed.
M-> $a

M-> parm dirw

M-> \f

M-> w

M-> q

bce (boot) 1211.5: M-> reinit

bce (boot) 1211.6: M-> boot star


Run Incremental Hierarchy Backups

While ideally, you should run both Volume and Hierarchy backups, and you should run complete, consolidated/catchup, and incremental backups, we recommend, at minimum, that you run an incremental hierarchical backup while your system is up.

To do this, once you boot your system, type the following at the Initializer console:

x inc <operator initials> <tape-label>

For example:

x inc ejs ih0001 

Here, I'm using the convention that tapes beginning with "ih" are "incremental hierarchy" backup tapes.

You will see the following output:


sc_command r bk start_dump sys_dirs "ejs" 1 60
sc_command r bk "ih0001"
Ready
M-> 
CONSOLE: RELEASED

 1215  bk   Begin at 03/22/16  1215.6 pst Tue
 1215  bk   
 1215  bk   >user_dir_dir
 1215  bk   
 1215  bk   Type primary dump tape label:   
 1215  bk   Mounting tape ih0001 for writing
1215.7  RCP: Attached tapa_01 for Backup.SysDaemon.z
1215.7  RCP: Note (tapa_01) - ih0001,sys
CONSOLE: ALERT
1215.7  RCP: Mount Reel ih0001 with ring on tapa_01 for Backup.SysDaemon.z
<ih0001.tap>
attachTape selected unit 1
TAPE: creating new file
CONSOLE: ALERT
1215.7  RCP: Authenticate tapa_01.  It has blank label .
M-> 

Authenticate your tape, by issuing the following command:

x auth tapa_01 ***

Substitute, of course, the correct tape drive if tapa_01 is not the one chosen for your tape mount.

You will see output similar to the following:

Ready
M-> 
CONSOLE: RELEASED

 1215  bk   Mounted blank volume "ih0001" on device tapa_01
 1215  bk   Begin at 03/22/16  1215.6 pst Tue
 1216  bk   Normal termination 03/22/16  1216.1 pst Tue.
 1216  bk   
 1216  bk   >system_control_1
 1216  bk   Begin at 03/22/16  1216.1 pst Tue
 1216  bk   Error file attached to file "start_dump.03/22/16.1216.ef".
 1216  bk   Normal termination 03/22/16  1216.3 pst Tue.
 1216  bk   
 1216  bk   >documentation
 1216  bk   Begin at 03/22/16  1216.3 pst Tue
 1216  bk   Normal termination 03/22/16  1216.4 pst Tue.
 1216  bk   
 1216  bk   >library_dir_dir
 1216  bk   Begin at 03/22/16  1216.4 pst Tue
 1216  bk   Normal termination 03/22/16  1216.4 pst Tue.
 1216  bk   
 1216  bk   >daemon_dir_dir
 1216  bk   Begin at 03/22/16  1216.4 pst Tue
 1216  bk   Normal termination 03/22/16  1216.5 pst Tue.
 1216  bk   
 1216  bk   >site
 1216  bk   Begin at 03/22/16  1216.5 pst Tue
 1216  bk   Normal termination 03/22/16  1216.5 pst Tue.
 1216  bk   
 1216  bk   >dumps
 1216  bk   Begin at 03/22/16  1216.5 pst Tue
 1216  bk   Normal termination 03/22/16  1216.5 pst Tue.
 1216  bk   
 1216  bk   Dump finished.
 1216  bk   Dumper going to sleep.
 1216  bk   
 1216  bk   r 12:16 36.335 2784
 1216  bk   
-->  bk

The incremental hierarchy backup daemon (Backup.SysDaemon) will wake up periodically and dump changed segments/directories to tape. When the current tape is filled up, the Initializer console will prompt for another tape.