[OpenAFS-devel] PATCH: break up cache into reasonable subdirs for large cache sizes

[Neulinger, Nathan]

I'll give you one reason why I don't like it. Reiserfs on 2.2.x. If you
enable this here, then a currently high-performance cache will be slower
than snot. Granted, that is due to interaction with reiserfs, but it works
great on 2.2.x right now.

I would tend to think that this would be slower on any filesystem that
handles large numbers of files in one dir well, but I don't have any
evidence to back that up.

I think that I'd be appeased (for the optional part) if two options were
available. Like we have -chunksize and -(whatever the other one is) - allow
specifying either number of files in cache dir, or number of subdirs. Then,
simply special case the numberofsubdirs==1 to skip the subdir part. That
would make it compatible/equivalent to existing cache. 

----

On the "deletion of old style cache" question - does it do the delete before
or after creation of the new subdirs+files? If after, that may cause issues,
cause it would mean temporarily using a lot larger number of inodes on the
disk. If you've got your inodes+partitions sized for the afs cache, that
might be a problem.

-- Nathan 

-----Original Message-----
From: joda+openafs@pdc.kth.se
To: Neulinger, Nathan
Cc: 'Derek Atkins '; 'openafs-devel@openafs.org '
Sent: 7/8/01 7:59 AM
Subject: Re: [OpenAFS-devel] PATCH: break up cache into reasonable subdirs
for large cache sizes

"Neulinger, Nathan" <nneul@umr.edu> writes:

> If this were made optional (i.e. add a param to afsd), I would have
> no objection to it being added in. Personally, I'd like to see the
> cache implementation be much more modular - to allow plugging in
> different back-end storage mechanisms in different cases.

I agree with Sam. If we feel that it's good enough, it should be the
only way to create cache entries, not something you can enable if you
like it.

/Johan