Mailing List Archive: Problems when PRImary hangs ...

Hi,

We had a power-supply failure on our primary machine a couple of days ago,
and when I tried to bring up the secondary, the FSCK showed heaps of problems.

Since we had a spare power supply, we quickly changed the power supply and
brought the primary back up, however, it is obvious that there are still
problems.

Since then I have added a nightly script on the secondary to do an fsck -n
of /dev/nb0, and it is clear that there are times when the buffer cache has
not been flushed on the primary that there are problems ...

If I run sync on the primary a couple of times and then do the fsck -n
/dev/nb0 on the seconday, it reports no errors, but without doing that, I
can get lots of errors.

I use the B protocol.

What can I do to minimize the possibility that the secondary will have
these problems?

Regards
-------
Richard Sharpe, sharpe@example.com
Samba (Team member, www.samba.org), Ethereal (Team member, www.zing.org)
Contributing author, SAMS Teach Yourself Samba in 24 Hours
Author, Special Edition, Using Samba

On Fri, 24 Nov 2000, Richard Sharpe wrote:

> Hi,
>
> We had a power-supply failure on our primary machine a couple of days ago,
> and when I tried to bring up the secondary, the FSCK showed heaps of problems.
>
> Since we had a spare power supply, we quickly changed the power supply and
> brought the primary back up, however, it is obvious that there are still
> problems.
>
> Since then I have added a nightly script on the secondary to do an fsck -n
> of /dev/nb0, and it is clear that there are times when the buffer cache has
> not been flushed on the primary that there are problems ...
>
> If I run sync on the primary a couple of times and then do the fsck -n
> /dev/nb0 on the seconday, it reports no errors, but without doing that, I
> can get lots of errors.
>
> I use the B protocol.
>
> What can I do to minimize the possibility that the secondary will have
> these problems?

1 - Journalling is the best solution.

2 - Use the "sync" mount option with ext2 (enables synchronous writes to
the filesystem, which are slower than async ones)

3 - Decrease the number of maximum dirty buffers (read
/usr/src/linux/Documentation/sysctl/vm.txt). With this you only minimize
the chance of having consistency problems, but there is no guarantee.