"Limits" Option Not Archving Instances In CE10

[horatiog]

Hi,

Running CE10 on Windows 2000.

We ran out of space on our production server yesterday. We are considering buying a bigger disk but in the meantime I urgently needed to get some space back.

Stopping & restarting the services and some judicious culling of spurious files on the disk yielded about 0.5GB. Not really a big enough comfort zone, so my next thought was archiving some of our instance history.

AFAICS, in CE10, you can apply "Limits" either globally (from "Settings"), or at a folder level (under the "Limits" tab) or at an individual object level (from the "Limits" button under the Object's History tab.) From the documentation I have concluded that the lower level "Limits" override the higher level ones.

I can see that a previous administrator had set a 40 day limit on instances for the "Everyone" group on the top level folder for one of our Divisions' reports hierarchy. (I assume this cascades down onto all the subfolders as well?) However, this was not set on our other Divisions' hierarchies. So, as a "quick win", I got authorisation to replicate this Limit on our entire folder hierarchy.

This seemed to do nothing initially. (Our "popular" reports still had thousands of entries stretching back several years.) I thought that the archive process might run in the background so left it overnight. When I got in this morning, some reports seem to have lost some of their instances but others still have a large intact history. (Free disk space is now at 1.5GB - still a bit tight!) My next thought was that for these reports were a long history has been retained, there were some lower folder or object specific overrides in place. However, a methodical search revealed no such overrides in place. My next thought was that CE may only archive an object when a new instance is generated for it, but running a report does not seem to do anything either.

So the bottom line here is, What am I doing wrong?

I know I can manually delete the history but this does not really help in the longterm and it will be a right PITA to go round all our 3-400 reports manually searching for and deleting old history!

Any help that can be offered will be gratefully received.

Regards,
Harry Haines
Reprots Analyst.

 

[rheard]

Harry,
did you get round this issue yet? I've been meaning to investigate why our instance history grows, but not got round to it yet - I'd be interested to hear any solutions... we're running CRServer 11

Russ

 

[horatiog]

I was speaking to some colleagues on Monday and they seem to think that history is only ever archived on a user by user basis, so the rules are only applied to a specific user's history for a specific report when that user runs that report. Our experience seems to be bearing this out in that we have now slowly clawed our way back to ~7GB free space. However, I haven't had time to prove this conclusively.

The way these rules are applied seems a bit strange to me, especially when the rule is "object" rather than "user" based. You would at least expect that it would delete all history over 40 days old for an object when anyone ran that report. Ideally you would expect a background process to be operating periodically to walk through all reports history, apply the net effect of any rules in place and carry out the relevant deletions.

 

[sandy3269]

I have been struggling with this issue for awhile now, more than a year, so here is the gist of my research.

Crystal Enterprise will keep your instances to a set limit, BUT only if you set that limit before there are active instances in the directory (From BO Support). Because we never did this globally, we started setting it at the top level folder by folder as we could, (different departments had issues regarding deleting all of their instances).

This seemed to work within those folders, the system started maintaining the set levels. But I didn't get the disk savings I thought I would.

I came up with the plan to query the CE metabase returning all the instance names, run a query of the FRS directories, then match them, removing those identified as Orphans. Turns out there is a bug in CE that stops queries from running longer than 9 minutes, which currently has me stumped (any help here?).

But looking at the query of the FRS directories shows thousands of more file directories than there are reports, and most of them empty.

My final solution when I'm able to will be to create a new metabase, set CE to use it, import folders and set limits, import all objects other than instances and basically start fresh.