Weekly edition	Kernel	Security	Distributions	Contact Us	Search
Archives	Calendar	Subscribe	Write for LWN	LWN.net FAQ	Sponsors

I'm sorry to see this abandoned

Posted Jun 18, 2009 22:26 UTC (Thu) by dlang (✭ supporter ✭, #313)
Parent article: What ever happened to chunkfs?

I was actually getting ready to track this down.

I don't see the need for a journal on each chunk as being a critical problem. It does mean that the plan of having 1G chunks doesn't work, but for many larger filesystems that was already questionable (I try to split my _files_ into 1G chunks if I can, but if they are < 10g I may not try too hard)

I have a use case where I want to have a ~140 TB array that will essentially be a single directory of files, I was thinking in terms of chunk sizes in the multi-TB size, at that point the need to have a separate journal per chunk isn't that bad in terms of overhead

yes, 1G chunks with many small files can be reasonable, but in most cases where you are talking extrememly large logical drives, you aren't dealing with text files, you are dealing with large files as well.

the fact that this wouldhave allowed filesystem checks to a chunk what the logical drive is online would have been a very useful thing.

(Log in to post comments)

I'm sorry to see this abandoned

Posted Jun 20, 2009 3:04 UTC (Sat) by vaurora (guest, #38407) [Link]

The code is out there. :)

I'm sorry to see this abandoned

Posted Jun 20, 2009 7:30 UTC (Sat) by dlang (✭ supporter ✭, #313) [Link]

yeah, but now instead of just encouraging the author to get it into the mainline (in part by being able to provide real-life use-case and production use) now I have to track down another programmer to learn the code and try to push it into mainline.

a much harder task (but still a possible one as you point out ;-)

Apr	MAY	Jun
	20
2014	2015	2016