On 12 4 2005 at 5:51 pm -0400, Steve Abrahamson wrote: >I have, and it doesn't seem to help too much. Granted, I get free space >back, but that's not nearly as important as performance. I'm waiting >seconds between click and response, especially when PM is fetching mail >at the same time (just best to take your hands off the keyboard and sit >back during that), and... well, that just shouldn't be happening.
Actually, I have been finding this a LOT with the 5.2 beta, although I wasn't previously, and I've been holding off saying anything about it on the assumption it may be a development artifact. Basically, I'll go "Connect..." and will get the SPOD for 10 to 20 seconds before anything happens. As I said though, with prior versions this hasn't been a problem for me (on a 3-year-old dual 867mhz G4). >I have some concern about my database and fragmentation. My message >database is about 500 megs (when I compact it it goes down to 200-250 >megs). But when I look at my disk fragmentation, my message database is >(before compacting), over 1,600 fragments. And I'm thinking, gee, maybe >PM's spending way too many cycles managing disk access. PM doesn't manage file fragments; you'll have to speak to the filesystem about that. :) How do you determine the fragmentation? I'd be curious (academically) to see how mine fares. Are you saying that doing a Compact doesn't result in a mostly unfragmented DB? >The question is what to do about that. The database is open all the time. >OS X 10.3's journaled file system should be able to defragment files, and >it seems that it does a reasonably decent job of that, but it only gets >to do that when the file is opening (I think). If the database is always >open, it's just going to keep writing and writing and writing, and it's >going to get hideously fragmented. The auto-defragmentation thing only happens for small files (something on the order of 10 MB I think but I forget offhand). Google the technote for details. >Wondering aloud, might there be some way for PM to encourage the OS to >keep it's database less fragmented? Some routine that poked the OS to >"maintain file <xyz>" during 3am OS housekeeping tasks? How about a cron job which copies the file to a temp file, deletes the original, and renames? If you have sufficient free space for the target file, it should end up mostly defragmented by virtue of recreating it. Or so I would expect. >Alternately, why am I experiencing a problem where others are not, what's >the difference, and what behavioral changes can I make here to match >those who are not having a problem? This is a better question -- I suspect that the fragmentation thing is a red herring. -b -- Ben Kennedy, chief magician zygoat creative technical services 613-228-3392 | 1-866-466-4628 http://www.zygoat.ca
