I should try that, we save copies of our F5 configs for backups, but sometimes 
I need to look through them to see what changed and when (now that I'm not the 
only one making changes on it), though its kind of a mess since its just a big 
directory of dated files. Plus it would probably be more space efficient, 
though if I moved the backup directory to the NAS then space wouldn't be an 
issue. 

The nightly backup is of both a ucs and the scf....the scf into revision 
control I think would be helpful...being ascii and all. While it would be a big 
harder with the gzip'd tar file with ucs extension. For now I think I'm the 
only one that makes changes outside of the GUI to the F5, including some that 
don't get into the ucs. They made it harder to add your own files to it...and 
there's no guarantee that when I upgrade they won't get ignored. That tripped 
me up the last time I upgraded the F5. Plus someday we'll need to upgrade to 
new units. Originally they said these would be the end of the line....though 
its probably more because when people's applications fail...they always blame 
the F5 for marking them down, or causing them to go down, etc. 

Like start of class rush slammed the student information system hard....I saw 
that the service was taking longer and longer to return to service checks, so I 
bumped out the timeout in the health monitor (to that recommended in the latest 
F5/peoplesoft guide). 12 hours later they made some change, and suddenly 
students are seeing other people's data. And, they blamed the F5. Wanted to 
know if it was caching or something. No, we don't have that enabled anywhere. 
Kept insisting that we must be caching somewhere to cause this problem. Didn't 
even know we had the feature. In the aftermath, they want all the F5/peoplesoft 
recommendations implemented. Which includes caching, compression and use of 
oneconnect. Well, we don't have a compression license...the free 5Mbps isn't 
going to cut it. But, the features they claimed was breaking they're 
application are ones they want turned on now. Though later it was revealed that 
the DBAs don't know how the web stuff works at all....but they'll play with its 
settings when they think they need playing with. And, turns out there was a 
peoplesoft bug that was causing the session overlaps. Even though the unit 
isn't EoSL, it is EOL...which apparently means we can't buy licenses to add 
functionality to it anymore. They want more SSL TPS, since using 2048bit keys 
cuts our 5000 TPS license to a 1000 TPS license. 
----- Original Message -----

> In a previous job, I had an epiphany that the most critical database
> that the company used was actually not that big. At close of
> business each day, I did a full text dump of that database and
> auto-committed it into svn. This gave us a history of the database
> more or less in perpetuity, with a daily granularity.

> The idea was to protect against a situation where some bad data or
> corruption crept into the database but didn't get discovered for
> many moons. (Given the state of the application that was feeding
> data in, this was not inconceivable) This would give us a way to go
> back and untangle things.

> --
> Christopher Manly
> Coordinator, Library Systems
> Cornell University Library Information Technologies
> [email protected]
> 607-255-3344
_______________________________________________
Discuss mailing list
[email protected]
https://lists.lopsa.org/cgi-bin/mailman/listinfo/discuss
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to