Linux sometimes kernel panics when starting memcached... (eek)

[Justin Matlock]

I haven't been able to duplicate this on 2.4.21 or 2.6test4, and I slam my
servers..

I'd say use the same kernel that LJ is using on their memcached servers --- 
they don't seem to have the KP problem.. :)

'course, I don't know what kernel they're running over there...  2.4.21
works great for me (2.6test4 seems to work, too, but that's pretty ambitious
for a production server).

J


----- Original Message ----- 
From: "Frank Precissi" <corvus@vadept.com>
To: <memcached@lists.danga.com>
Sent: Tuesday, September 02, 2003 9:23 PM
Subject: Re: Linux sometimes kernel panics when starting memcached... (eek)


> Are there any test scripts written so we can load our memcaches boxen to
see
> if the'll panic the kernel before they are put into production? :)
>
> Frank
>
> On Tuesday 02 September 2003 02:25 pm, Brad Fitzpatrick scribbled on
digital
> paper:
> > He said he would, if taral would send him updated patches, which he
> > hadn't.
> >
> > rtsig scales the same way as epoll (that is, with the number of active
> > connections, not total connections), but is a fraction slower.
> >
> > On Wed, 3 Sep 2003, Anatoly Vorobey wrote:
> > > On Tue, Sep 02, 2003 at 02:06:58PM -0700, Brad Fitzpatrick wrote:
> > > > I'm wondering if we should ditch epoll and libevent and do our own
> > > > abstraction, choosing between either kqueue on FreeBSD or real-time
> > > > signals on Linux.
> > > >
> > > > Avva?
> > >
> > > I don't know. I've never tried real-time signals. Are they as good
> > > as epoll, speed-wise? (epoll is pretty good).
> > >
> > > Also, I recall you were trying to get the libevent guy to accept
> > > the real-time signals patch, anything new on that front?
> > >
> > > --
> > > avva
>
> -- 
> People who take cat naps don't usually sleep in a cat's cradle.
>
>