| Load Reviews-Backup and Contingency |
We identified the need for a redundancy in the load review software in the case of disk failure, server issues and other things that could prevent us from running the load review. Since this is a mission critical task, we originally created a whole new set of scripts to mirror the original scripts and set up new areas.
In all cases, we have since consolidated the original and _bak versions of the scripts. The unified code will select directories to work with based on calls to lr_suffix.pl (which returns "-bak" if the current host is colossus-v, and a null string otherwise), or to "lr_suffix.pl -x" (which returns "_bak" or a null string.)
For the present, the three top-level lr scripts continue to bear _bak names on the OCC LAN: lr_bak, ACE-update_bak.pl, and history-files_bak.pl. Each is a soft link to code identical to the HEAD LAN version with no _bak subscript.
% setloads_bak
Then you would execute the lr_bak script the same way you would execute the lr script. The scripts have the proper location of the scripts coded in. Here's a sample session.
% su - acisdude
% setloads_bak
% lr_bak SEP2004A SEP1304
When the load has been approved, you MUST run ACE_update.pl -b even if the regular disks are back up. This will appropriately move data to the regular disks if they are available.
WARNING: The post approval scripts ACE-update_bak.pl and
history-files_bak.pl are ancient
versions, unlikely to work on the OCC LAN. Wait for the HEAD LAN to
become accessible again. Then make the appropriate file copies from
OCC to HEAD LAN
[TBD: documentation still required here], and run the relevant approval
script there.
To access another OCAT server:
For example: for the week of September 20 2004 A load,
For example: for the week of September 20 2004 A load,
Last updated:
01/27/16
What if the OCAT is down?
If you discover that the OCAT is not working, for example, you see
many server error messages from the lr or lr_bak code, you should do
the following:
You DO NOT need to run the backup code if this is the only problem. LR
has been modified to allow a server name to be passed through.
% lr -s server_name current_load last_week_load
OR
% lr_bak -s server_name current_load last_week_load
I can't find the backstop file on Lucky.
Occasionally, the link between the OCC and Garden street will go
down. Sometimes we know about it, other times, it is unexpected. We
may not be able to access the machine Lucky. We need to get the
backstop.tar.gz file from Lucky to the HEAD LAN to execute the load
review.
To work around this, you can do the following:
Please note that this works for both the lr and lr_bak codes. Just
make sure you copy the backstop.tar.gz file into the correct
directory.
% cp /some/directory/that/FOTMP/told/me/SEP2004A.backstop.tar.gz
/data/acis/LoadReviews/2004/SEP2004/oflsa/FOT MP forgot the CRM file. Now what?
In hurried cases, the CRM file can be forgotten in the backstop tar
file. This is usually in a replan situation and we have changed the
lr --break.pl code to deal with this. However, this could happen when
you are running lr and you need to get the CRM file in there.
To work around this, you can do the following:
% cp /some/directory/that/FOTMP/told/me/SEP2004_CRM_Pad_SI1.txt
/data/acis/LoadReviews/2004/SEP2004/oflsa/When do I run the backup scripts?
Here is a short list of situations when the backup scripts should executed:
ACIS Monitor
Last modified: Fri Feb 14 14:59:00 EST 2014