David O'Brien wrote: > On Fri, Apr 30, 2004 at 02:30:40PM +0200, Thomas Quinot wrote: > >>* Panagiotis Astithas, 2004-04-30 : >> >> >>>I was wondering (since being bitten by this occasionally) why don't >>>we change the default dumpdir in /etc/default/rc.conf to /usr/crash >>>instead of /var/crash? The default partitioning scheme in sysinstall >>>(when you press 'A') creates a /var with only 256MB, whereas it >>>retains a huge amount of space for /usr. Shouldn't we help Joe "the >>>defaults" User have working crashdumps? >> >>The proper fix would probably be to change the default partitioning >>scheme, not to move the crash dumps. I think one property we try to >>guarantee is that /usr be mountable read-only through NFS for a cluster >>of workstations, whereas /var is always mounted read-write, for its >>purpose is to contain files whose contents *vary* over time. > > > All correct. > > Panagiotis, please instead propose a proper size for /var. Hmmm, tricky. In my office alone I can count at least 5 systems with different amounts of memory, from 128 MB to 1 GB. Now only the first one can get crashdumps in a 256 MB /var. If we resize /var to 1 GB (as I usually do) we get working crashdumps in 4 out of 5. Upping it to 2 GB we get all 5 of my systems. The maximum configuration of 4 GB (for 32-bit systems) needs a 4.5 GB /var, which is not too much for the 60 GB hard disks we buy nowadays in laptops (servers come with larger disks and need more space in /var for logs, anyway). If we cater for 64-bit systems too, we need even more... I think I would be conservative and suggest a 1 GB /var for now. -- Panagiotis Astithas Electrical & Computer Engineer, PhD Network Management Center National Technical University of Athens, GreeceReceived on Sun May 02 2004 - 04:32:41 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:52 UTC