Yes, we suspected something like that... we have already increased innodb_buffer_pool_size from 32G to 64G (and have new DB nodes on the way) but it didn't help. There aren't dedicated DB nodes though.
We assumed it must be some tipping point thing, hence looking into purging. But like in the ticket, we don't see a big increase in memory usage during the query, nor are they using all the 64G available to them (or the 32 before that). Many thanks, Luke -- Luke Sudbery Architecture, Infrastructure and Systems Advanced Research Computing, IT Services Room 132, Computer Centre G5, Elms Road Please note I don’t work on Monday. > -----Original Message----- > From: ahmet.mer...@uhem.itu.edu.tr <ahmet.mer...@uhem.itu.edu.tr> > Sent: 23 February 2021 14:20 > To: Slurm User Community List <slurm-users@lists.schedmd.com>; Luke > Sudbery (Advanced Research Computing) <l.r.sudb...@bham.ac.uk> > Subject: Re: [slurm-users] Slurmdbd purge settings > > Hi; > > May be the database can not fit innodb buffer any more. If there are > enough room to increase this value(innodb_buffer_pool_size) , to find > reason, you can try the increase. > > Ahmet M. > > > 23.02.2021 17:03 tarihinde Luke Sudbery yazdı: > > That great, thanks. We were thinking about staging it like that, and > using days is simpler to trigger than waiting for the month. > > > > We will also need to increase innodb_lock_wait_timeout first so we > don't hit the problems described in > https://bugs.schedmd.com/show_bug.cgi?id=4295. > > > > Anyone know why sreport would suddenly so much longer in the first > place, though? > > > > Many thanks, > > > > Luke > >
