Hi all,
when upgrading the BDII to version 3.2.x one must ensure there are _no_
lcg-bdii-update cron jobs left!
The new BDII has an update process running continuously.
Any cron job will eventually cause the update process to die,
making the BDII serve stale data.
Look for such cron jobs in /etc/cron.d and in the output of "crontab -l"
for root.
On Mon, 4 Apr 2005, Laurence wrote:
> Here are CERN we run 7 production BDIIs. We have not seen this error
> and haven't had to do any maintenance on the BDIIs. The stale slapd
> problem is more with MDS than the BDII as the MDS script doesn't have a
> very good shutdown method. Sometimes a slapd thread serving a query can
> hang and will not be shutdown correctly. The BDII has a better shutdown
> method but there could always be some unusual situation in which this
> will not work.
> The current version of the BDII does have a fews weaknesses. The
> database swap method is not very good and the reliance on the cron job
> can cause timing problems. For example if you restart the BDII at the
> same time as the cron job runs, you can get into difficulties.
> All of these problems have been address in the latest version of the BDII.
>
> Laurence
>
> Lev Shamardin wrote:
>
> > On Mon, Apr 04, 2005 at 10:39:04AM +0200, Laurence wrote:
> > > I am not aware of any hanging problems with this version of the BDII.
> > > The only thing that I can think of is that there is a stale slapd
> > > process hanging around. Try to stop the BDII, kill all slapd processes
> > > and restart the BDII.
> >
> > This *surely* helps, but what causes this stale process to appear on a
> > BDII
> > not touched by anyone, and what should I do in this case? Should I just
> > restart the BDII every N days/hours/minutes? What is the best value
> > for N?
> >
> > --
> > Lev.
> >
>
|