RecoverFromCacheDoc.tioga
Last edited by Bob Hagmann, May 15, 1986 8:11:48 am PDT
RecoverFromCacheDoc
CEDAR 6.1 — FOR INTERNAL XEROX USE ONLY
Recover From Cache Documentation
Bob Hagmann
© Copyright 1985 Xerox Corporation. All rights reserved.
Restore files to a file server from the cached copies on the workstation. Files can be either copied or listed.
Keywords: FS cache, File Server Disaster
XEROX Xerox Corporation
Palo Alto Research Center
3333 Coyote Hill Road
Palo Alto, California 94304
For Internal Xerox Use Only
1. Usage
RecoverFromCache {-c} {-l} Volume {Server}
The volume must be a Cedar volume. The normal one is named 'Cedar'.
The 'c' switch means auto confirm. Normally, you are prompted before each file store.
The 'l' switch means list only.
The Server, if specified, restricts the search to files from that Server. Use Pseudo-server names.
Examples:
RecoverFromCache -l Cedar Indigo
List the files in the cache on volume Cedar from file server Indigo that are not on the server.
RecoverFromCache -l Cedar Cedar
List the files in the cache on volume Cedar from pseudo server Cedar that are not on the server.
RecoverFromCache -c Cedar Cedar
Copy the files in the cache on volume Cedar from pseudo server Cedar that are not on the server. No confirmation needed
2. Discussion
First of all, I'm sorry you are reading this. You have had something bad happen, and you need help. However, no warranty is expressed or implied by this program.
What this program will do for you is write onto a file server files in your FS cache that are not (now) on the file server. This will help recover from a file server crash, and also from inadvertent deletion of files (via Chat probably) on the file server where there is a copy of the file in the cache on the workstation.
You have to know the Cedar volume where the cache is. This is probably named 'Cedar'.
You also have to know the file server that you want to recover. You must use the Pseudo-server name if that is how the files are cached. And you must have the Pseudo-servers set up correctly.
This program has two phases. It first enumerates the local disk to find create dates (an optimization). This takes about 3 minutes on a Dorado with a T-80. Next, it looks at files with the highest version number for the file server specified (or all cached files if no server), and sees if there is a file with the correct create date (version numbers are ignored!) on the server. If so, all is OK and we go on to the next file. If not, then the file is listed if the l switch is set, copied if the c switch is set, or you are asked if you want to copy it otherwise. If you respond "y" followed by a return, it will copy it. "n" will not copy. In any case, the enumeration continues.
With a fairly idle IFS, it took about 4 minutes to look at 482 files. A T-80 typically caches 2,000 files. During the second enumeration, the IFS is quite busy.
I suggest you use the l switch first to see all the files that would be written first. You may discover gobs of old files from obsolete directories. You can figure out how much trouble you are in.
Warning: this program strains the file server. Be considerate.
Good luck.