Boston Linux & UNIX was originally founded in 1994 as part of The Boston Computer Society. We meet on the third Wednesday of each month at the Massachusetts Institute of Technology, in Building E51.

BLU Discuss list archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Discuss] Gluster startup, small-files performance



F. O. Ozbek wrote:
> I mean, come on, look at the claim you are making:
> " MooseFS isn't reliable", have you ever done any tests?

I know that if the power suddenly fails before data has been written to
non-volatile storage then that data will be lost. That's how computers
work (or don't work as the case may be). While I haven't tested power
fault conditions with MooseFS specifically I have seen data loss and
file system corruption happen with many other file systems -- and had to
clean up the messes afterwards. On the flip side, I have tested power
fault conditions under working fsync and O_SYNC conditions and have been
spared the worst of the post-test cleanup. Using these experiences I can
extrapolate how MooseFS will behave under similar conditions. It won't
be 100% accurate but I don't need 100% accuracy to easily identify some
failure conditions.

It seems to me that you've done your homework for your own installation
and covered the most likely failure modes as best as you can. That's not
a bad thing. What I think is a bad thing is relying on backup power as a
primary means of preventing data loss. Backup power has failed
catastrophically for Amazon in at least three different data centers in
three different locations in the past few years. Can you be sure that
power won't fail just as catastrophically for you? Of course you can't.
Nobody can. Thus my caution about using MooseFS for important data.

-- 
Rich P.



BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!



Boston Linux & Unix / webmaster@blu.org