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

Re: [request-tracker-maintainers] Bug#595054: request-tracker3.8: Race condition between RT3.8+apache2 and MySQL when booting by insserv



On Tue, Aug 31, 2010 at 11:50:31PM +0200, Salvo Tomaselli wrote:
> On Tuesday 31 August 2010 22:39:18 Dominic Hargreaves wrote:
> > * Arrange for database servers to start before Apache
> I wouldn't like that. We would be removing possible parallelism (in a far far 
> away future when we would be using this) at boot.

In other words, although the notion of a sequencing is unlikely
to result in functional problems (eg the database server failing to start
up because it needs to talk to Apache) this doesn't go fo possible quantative
problems (slower boot time) which the parallel booting scheme is trying
to address. I suppose I hadn't fully considered this aspect. I agree that
we should avoid slowing down the boot process as a workaround for a buggy
applicaton.

> > * Arrange for RT to be more robust when a connection to the database fails
> That sounds better.
> Sounds to me that we assume the SQL server used is going to be on localhost. 
> It's quite a strong assumption and a connection can always fail, so the 
> application shouldn't crash.

Yes. I should have made more clear that I wouldn't intend to disregard a
fix in the application; it was a question of the possibility of a pragmatic
fix in the short term which I wanted to discuss.

> The original email was talking about a manual restart of apache. I didn't try 
> it myself but... does the apache server terminates because of a failed 
> connection to a database from a web application running on apache itself?
> _If_ this is the case, this behavior is absolutely wrong and should be 
> considered a bug without trying to look for workarounds.

Yes. Perhaps there is also a separate bug to be filed against Apache,
although that tight binding between a mod_perl application and Apache
itself means that perhaps this would be considered NOTABUG, and something
which the application should be relied upon to get right.

Thanks for your feedback.

Dominic.

-- 
Dominic Hargreaves | http://www.larted.org.uk/~dom/
PGP key 5178E2A5 from the.earth.li (keyserver,web,email)


Reply to: