You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Ben Collins-Sussman <su...@newton.collab.net> on 2000/11/13 16:54:47 UTC
apr repository!
Hey, I can update my apr/ directory really fast now; it used to take
forever. Is this because Brian has somehow fixed a caching problem on
www.apache.org? Is apr's repository living on a new box, now that
it's an independent ASF project?
Just curious.
Re: apr repository!
Posted by Brian Behlendorf <br...@collab.net>.
On 13 Nov 2000, Ben Collins-Sussman wrote:
> Brian Behlendorf <br...@collab.net> writes:
>
> > I put /tmp on an MFS, since cvs pserver uses /tmp pretty
> > heavily.
>
> MFS == Memory File System? Is this some kind of RAM disk?
yes. since /tmp is definitely not required to survive a reboot, and we're
OK on memory usage (aside from CVS clients like the one I posted about =)
it seems safe. Looks like CVS makes an entire copy of what it wants to
send to a client to a dir in /tmp before sending it out; makes a bare
minimum of sense if you have reads locking writes to the repository, so
it's not locked while someone does a big long read-only update.
Brian
Re: apr repository!
Posted by Ben Collins-Sussman <su...@newton.collab.net>.
Brian Behlendorf <br...@collab.net> writes:
> I put /tmp on an MFS, since cvs pserver uses /tmp pretty
> heavily.
MFS == Memory File System? Is this some kind of RAM disk?
Re: apr repository!
Posted by Brian Behlendorf <br...@collab.net>.
On 13 Nov 2000, Ben Collins-Sussman wrote:
> Hey, I can update my apr/ directory really fast now; it used to take
> forever. Is this because Brian has somehow fixed a caching problem on
> www.apache.org?
I put /tmp on an MFS, since cvs pserver uses /tmp pretty
heavily. There was one other change to how directory lookups are
cached that may have also helped. Glad to see it made an actual
difference!
Brian