[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: slapadd filching memory - NOT

Jon wrote:
In my testing for this and other problems I tried 2, 4, 8, and 16 Mb. I'm happy to boost efficiency anywhere, but my systems spend more time as a server than in loading data, so I'd prefer to tailor the cache settings to a live profile. I'll only be loading this data this big every so often, and I'm just as happy to cut it into chunks. What I'm more concerned about is the memory staying gone...

Jong wrote:
You can change the DB cache size after loading data.

run db_recover
change cache size setting in DB_CONFIG
run slapd

Thank you very much for your suggestions. I'm definitely still learning.

Howard Chu's recent note on cache size estimates gives me a better idea what I need. I recently tried setting 100Mb cache in a single pool for the same load. It took 12 hours to create a ~500 Mb database. Perhaps I should try a full 500 Mb next time (I have a Gb)? At least for loading.

As both you and Andrew Findlay (thanks) correctly pointed out, memory was NOT the issue with my slapadd process. I must've had something else happen simultaneously. Running it again, I am left with a stable system (albeit, the next day). It performs nicely in action.

Jon Roberts