just to share the experience : we got the problem with a CF used as system in an embeded box for a wearable camera, the script was rewriting the same jpg file every 5 sec, after a few days it got "burned" and led to writing error on that file... If I remember well we could get the card back in order by reformating it (and then reallocating the block to living parts) but it can happen... I suppose it could be a problem with some log file f.ex.... maybe something to pay attention to. Appart from that, I use a laptop with Xubuntu in which I swaped the HD by a 4GB CF with IDE adapter and no problem until now...
Ol. (now preparing debian systems for testing amd64 video compression speedup with kernel64 ... ) Aymeric Mansoux wrote: > [email protected] said : > >> Inidentally, I just recalled coming across something regarding read/write >> limitations of USB keys before they die, but maybe this has been resolved >> enough for P:D t(ça aurais été pas mal d'avoir des info sur ça avant, à la >> dernière >> palabre, on a pas développé la question très loin, vu le manque >> d'information sur les points à relier - photos, etc ... - à peine j'y >> ai découvert la topologie envisagée avec une vague mention sur le role >> de chaque noeud... encore un projet "à l'arrache" ... :/ )o be viable on USB? >> In any case, I'll try to take a read at the P:D archive. >> > > lots of myths and legend on this ... and also unknown facts. > > Of course wear and tear is happening earlier on Flash memory, but there > are no solid benchmark on this to my knowledge. Also recent flash > memory chip controlers are able to "smartly" predict upcoming tearing, > and/or reallocate dead blocks to extra non allocated space, and/or > spread writing data over all the blocks to avoid hammering the same over > and over. > > Unfortunately most of these techniques are heavily patented, so it's > hard to know what's effective and what's marketting. > > One thing for sure is that, unless you write a software that is going to > hammer your memory and write nonstop on it, it's unlikely that you'll > reach the maximum cycles with a normal usage, unless you have a flash > memory that has a sloppy or buggy wear levelling/balancing, or simply > not meant for that. I would assume you are safe with todays' commercial > SD,CF,USB keys, etc ... It's another story for rewritable Flash in > some microcontrollers. > > I used to be cautious about this though, but I tend to care less > nowadays as I never experienced any problems with that. > > For extra safety you could always make some system partitions read-only > and avoid using journaled FS. but ... not sure it's really worth the > trouble... ;) > > a. > > > > --- > [email protected] > irc.goto10.org #pure:dyne > > --- [email protected] irc.goto10.org #pure:dyne
