Technical article 116887

OpenServer 5, The verification of a BackupEdge backup takes almost 24 hours.

OpenServer 5.0.5OpenServer 5.0.6

Written · Last revised

Problem

The verification of a BackupEdge backup takes almost 24 hours.

Using the "/etc/edge.nightly" cron, configured by "/etc/edge.cronset", backing up 8GB approximately to DLT tape takes approximately one hour. However, the verification process takes almost 24 hours to complete.

Cause

Indexing under v01.01.08 can take a very long time depending on the
file names present on the system.  Directories that contain a large
number (thousands) of files can exceed the indexer's cache, causing
the index to perform extremely slowly.

By default, BackupEdge's "edge.nightly" will verify and index the tape for the Fast Restore process.

An extract from the BackupEdge manual states:

If you are going to index a tape, it is strongly recommended that you also perform a Bit-Level Verification on the tape, since it does not take significantly more time than an index alone.

This problem has been reported to MicroLite Support and has been resolved in:

v01.02.00 Build 5

Solution

BackupEdge, a product available from MicroLite, at:

http://www.microlite.com

is a product used for backup and emergency recovery and is not supported by SCO.

It is recommended to run the "Manual Check" of the tape drive to ensure BackupEdge is reading at optimum efficiency.

To do this, run:

/usr/bin/edgemenu -> Admin -> Define Resources -> Select your Tape Drive -> Manual Check

This affects Fast File Restore performance only.  Specifically, it
measures the overhead of using the tape drive's fast-positioning
method versus reading and discarding unwanted data.  Its goal is to
measure the overhead of switching into fast-positioning mode, so that
when it is skipping from file to file during a restore, it knows for
each skip whether which mode will be quicker.

In order, to remove the indexing option follow these steps:

- Run "/etc/edge.cronset" and remove the INDEX option from the
  backup run.  This will remove the "-x" option from the cron entry.

a) In v01.01.08, "edge.cronset" parses the crontab file. Ensure that the line is not already commented out in the cron file using the command:

crontab -l
otherwise "edge.cronset" will fail.

If it is using the following commands to modify the cron to remove the comment:

SCO recommends taking a copy of your cron files first:

mkdir -p /local/cron
cp /usr/spool/cron/crontabs/* /local/cron

Using your favorite editor modify the root file. Remove the "#" from the /etc/edge.nightly line.

Save the file and make live with the command:

crontab /local/cron/root

b) In v01.02.00 and later, the crontab command line is entirely different because of advances in the scheduler. The indexing commands are stored in a per-backup-job configuration file that is designed to be user-editable if desired.

To test, run the /etc/edge.nightly command manually
For v01.01.08:

It can be run as it appears in the crontab (crontab -l).

For v01.02.00:
/etc/edge.nightly -H (job name)
For example:
/etc/edge.nightly -H simple_job

The Basic Schedule is the equivalent of the schedule one sees with "edge.cronset" in v01.01.08.

If the process just returns straight to the prompt, then there may be
a lock file present preventing the backup from continuing.  Check
root mail for any error message or alternatively remove the lock file:
For v01.01.08:
/usr/lib/edge/lists/edge.progress file.

The lock file is cleared when /etc/rc2.d/S88edge is started at boot time.

For v01.02.00:
rm /usr/lib/edge/lists/(job name)/schedule.lck
For example:
rm /usr/lib/edge/lists/simple_job/schedule.lck
for the Basic Schedule.  01.02.00 will send email notification if it
encounters a lockfile describing the problem, unlike 01.01.08.
NOTE:     Other options to /etc/edge.nightly include:
-m      (No Verification)
-M      (Verify but not Bit-Level) - Default
-n      (User to mail summary to)
-d      (Output file)
NOTE:     If your tape drive is attached to a "cha" Compaq SCSI-2 Controller
          and you are getting the following notices appear in your
          /usr/adm/syslog and messages files and on the console, but the
          backup and restoring is fine:

NOTICE: cha: SCSI data underrun, ha=x id=x lun=x cmd=x blk=x len=x Data arriving from a SCSI data peripheral under-filled storage (chaN07)

can be removed by changing the "int cha_print_notices = -1" to "= 0" in /etc/conf/pack.d/ida/space.c then re-linking the kernel with the command:

/etc/conf/cf.d/link_unix -y

This problem is likely to come from using BackupEdge as running a simple backup using "cpio" is fine.

This has been seen with Compaq Extended Feature Supplement (EFS) v5.36.

See also

For more information on DLT technology visit:

http://www.quantum.com/Service+and+Support/Default.htm

ta114321, "I need to restore my data from a Compaq server running with the Smart Array 3200 (ida) to another Compaq server running with the Smart Array 431 (clad) using BackupEdge."

ta115746, "Where can I find a simple backup script to backup my server onto a tape on a nightly basis?"

http://www.microlite.com/support/edgefax51.ts.html
http://www.microlite.com/supportpages/

From the Xinuos technical article database, record 116887. Reproduced with permission. The original is at www.sco.com/ta/116887, which is reachable but blocked from search engines.

Next step

Need this fixed rather than explained?

Tell us the release, what the system runs and what it is doing. We will reply by email with what we would do next.

  • You deal with the people who do the work
  • No obligation, and no upgrade required before we will help
  • We tell you plainly if something is not worth doing

Send an enquiry

We reply by email. If a server is down, say so in the first line. You can also write to help@scoexpert.com.