[HyperDB] Re: HyperDB Digest, Vol 6, Issue 3

Thu Feb 7 12:56:09 GMT 2008

The 32k limit is for subdirectories within a directory, not files within a
directory.
In the past we've had up to hundreds of thousands of files in a mysql db
directory. It's not recommended for performance reasons, but it still does
work.
Quentin
http://unblog.fr

On Feb 7, 2008 1:00 PM, <hyperdb-request at lists.automattic.com> wrote:

> Send HyperDB mailing list submissions to
>        hyperdb at lists.automattic.com
>
> To subscribe or unsubscribe via the World Wide Web, visit
>        http://lists.automattic.com/mailman/listinfo/hyperdb
> or, via email, send a message with subject or body 'help' to
>        hyperdb-request at lists.automattic.com
>
> You can reach the person managing the list at
>        hyperdb-owner at lists.automattic.com
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of HyperDB digest..."
>
>
> Today's Topics:
>
>   1. Re: documentation? (Dan Milward)
>   2. Re: documentation? (Matt Mullenweg)
>   3. Re: documentation? (King of punk)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Thu, 07 Feb 2008 12:21:01 +1300
> From: Dan Milward <dan at instinct.co.nz>
> Subject: Re: [HyperDB] documentation?
> To: hyperdb at lists.automattic.com
> Message-ID: <47AA40DD.5010106 at instinct.co.nz>
> Content-Type: text/plain; charset=UTF-8; format=flowed
>
> We are using the ext3 file system which apparently supports around
> 32,000 files in one directory.
>
> As we are using WordPress MU which creates multiple tables per user
> (more then normal as we are using several plugins). Because each of
> those tables is actually several files in the MySQL directory we have
> calculated that unless MySQL does something smart like nesting of
> directories to work around the problem...
>
> ...Then we will probably hit the limit of files at around 350 users. At
> which point we will not be able to create files in that directory which
> means no more users will be able to sign up with other possible nasty
> side effects for whatever user puts it over the limit - for instance
> that user might only get half tables created necessary for their
> wordpress blog to run.
>
> So what I want to know is whether or not MySQL does something smart to
> work around this - and if not does a separate machine running HyperDB
> sound like a good idea.
>
> And does HyperDB work well with Super Cache 2 :))
>
> d
>
> King of punk wrote:
> > Hello
> > I am also putting myself on that big MU adventures, but how come that
> > your server "freak out" with only 350 users ???
> >
> > Kingofpunk
> >
> >
> >
> >
> >
> > Dan Milward wrote:
> >> Hi Guys,
> >>
> >> We're mucking around with an MU powered site that we expect to get
> >> pretty big quite fast - we made the mistake of using an ext3 file
> >> system. So what this means is that when we hit around the 350 users
> >> mark our server will freak out due to ext3 limitations.
> >>
> >> So this seems to me like a good opportunity to use get a new server
> >> for just the databases and use Hyper DB on a new XFS filesystem.
> >>
> >> One thing I want to do though is use Super Cache 2. Does this sound
> >> like a good idea - can anybody see any reason why it wouldn't work.
> >>
> >> Ciao,
> >>
> >> Dan
> >> _______________________________________________
> >> HyperDB mailing list
> >> HyperDB at lists.automattic.com
> >> http://lists.automattic.com/mailman/listinfo/hyperdb
> >>
> >
> > _______________________________________________
> > HyperDB mailing list
> > HyperDB at lists.automattic.com
> > http://lists.automattic.com/mailman/listinfo/hyperdb
> >
>
>
> ------------------------------
>
> Message: 2
> Date: Wed, 06 Feb 2008 16:28:27 -0800
> From: Matt Mullenweg <m at mullenweg.com>
> Subject: Re: [HyperDB] documentation?
> To: hyperdb at lists.automattic.com
> Message-ID: <47AA50AB.4080905 at mullenweg.com>
> Content-Type: text/plain; charset=UTF-8; format=flowed
>
> Dan Milward wrote:
> > ...Then we will probably hit the limit of files at around 350 users.
>
> 32,000 / 3 files per table = 10666 tables / 350 blogs = 30 tables per
> blog? That seems high, even with plugins.
>
> Also as far as I know the 32k file limit is only with 2.4 kernels, not
> 2.6 ones.
>
> > So what I want to know is whether or not MySQL does something smart to
> > work around this - and if not does a separate machine running HyperDB
> > sound like a good idea.
>
> With InnoDB all tables are put in a single file.
>
> But really you should use a kernel made within the past few years.
>
> > And does HyperDB work well with Super Cache 2 :))
>
> I don't see how they're related.
>
> --
> Matt Mullenweg
> http://ma.tt | http://automattic.com
>
>
> ------------------------------
>
> Message: 3
> Date: Thu, 07 Feb 2008 01:48:02 +0100
> From: King of punk <kingattitude at gmail.com>
> Subject: Re: [HyperDB] documentation?
> To: hyperdb at lists.automattic.com
> Message-ID: <47AA5542.9040703 at gmail.com>
> Content-Type: text/plain; charset=UTF-8; format=flowed
>
> Maybe the solution is a ReiserFs linux partitioning ?
>
> Matt Mullenweg wrote:
> > Dan Milward wrote:
> >> ...Then we will probably hit the limit of files at around 350 users.
> >
> > 32,000 / 3 files per table = 10666 tables / 350 blogs = 30 tables per
> > blog? That seems high, even with plugins.
> >
> > Also as far as I know the 32k file limit is only with 2.4 kernels, not
> > 2.6 ones.
> >
> >> So what I want to know is whether or not MySQL does something smart to
> >> work around this - and if not does a separate machine running HyperDB
> >> sound like a good idea.
> >
> > With InnoDB all tables are put in a single file.
> >
> > But really you should use a kernel made within the past few years.
> >
> >> And does HyperDB work well with Super Cache 2 :))
> >
> > I don't see how they're related.
> >
>
>
>
> ------------------------------
>
> _______________________________________________
> HyperDB mailing list
> HyperDB at lists.automattic.com
> http://lists.automattic.com/mailman/listinfo/hyperdb
>
>
> End of HyperDB Digest, Vol 6, Issue 3
> *************************************
>