On Sunday 27 June 2004 15:20, Luke Kearney wrote:
> On Sun, 27 Jun 2004 13:48:07 +0400
> "Kentucky Mandeloid Mo." <[EMAIL PROTECTED]> spake thus:
> > I'm looking for file replication solution for FreeBSD.
> >
> > I have a task to replicate about 11G of files with 500k total files
> > count. I say replication not mirroring i.e. I need to copy modified files
> > to slaves almost imidiately after the change.
> > Tried to find some ready solution but fails.
> > There is FAM from SGI but it fails to work properly on FreeBSD (it makes
> > silly stat(2) calls).
> > kevent(2) also don't have anough API to watch >= 500k files.
> > So I see the only way is to have kernel module that watch kernel calls
> > for file operations.
> >
> > So is there some other points to build a file replication?
> Can I suggest you investigate rsync ? It will allow you to replicate
> file trees pretty much on demand across a LAN or indeed the internet
> should you choose to. Works over SSH so it's nice and secure and if you
> need virtually instant replication you could conceivably run this from a
> cron job every N minutes.
> /usr/ports/net/rsync

Yes. I use rsync already and its a great tool.
But actually its too slow with numbers of fles > 100k.
Its takes more then 5mins to found just few files to update in list of 500k 

So I decide rsync to be a transport manager for my own replication tool. 
Right now I'm writing a kernel module that catches all file syscalls that can 
modify a file. Its almost done exept rmdir & unlink calls - I don't know how 
to get a cwd from a thread. Function kern___getcwd sometimes fails with "not 
a directory" error. 
BTW, whats a good place where can I ask about kernel programming?

[EMAIL PROTECTED] mailing list
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Reply via email to