Child pages
  • Archiving Data from opsimX Machines
Skip to end of metadata
Go to start of metadata

Objective

There are 8 production machines (opsimX, where X=1 to 8) that have been replaced by 2 faster production machines (opsimblitz1 & opsimblitz2). There are working files (for multiple users) and multiple databases with simulations from pre-OpSim v3.0 versions of the simulator.  The objective of this page is to identify solutions to preserving this data in an accessible way so that the older hardware can be retired before it fails.

 

Proposal

The proposed solution is to aquire a 2Tb drive, mount it on opsimblitz2 (the Linux production machine) and simply copy all the data to this drive.

Here is a table itemizing the directories that are targeted for relocation (and preservation).

 

machine

/home/* (Gb)

/lsst/* (Gb)

 

opsim1

19

194

 

opsim2

30

213

 

opsim3

66

240

 

opsim4

21

176

 

opsim5

30

159

 

opsim6

6.5

79

 

opsim1:/backup/buffalo/opsim7

12

33

 

opsim8

15

89

 

Total

200

1183

1383 Gb

 

To save time, simply copying these directories to a new directory structure on the new drive will accomplish the goal without requiring detailed inspection of and decision making about every version of the code and output files be made.

The new directory structure might look like:

/archive/machinename/home/*

/archive/machinename/lsst/*

etc....

To access any of the databases from opsimblitz2, you would log in as yourself, and edit the file

/etc/my.cnf

to change the line

datadir=pathname_to_database 

A backup strategy will be needed - either add to current backup systems, or acquire a mirror disk.

  • No labels