On Thu, Apr 8, 2010 at 10:42 AM, Kevin D. Clark <kevin_d_cl...@comcast.net>wrote:
> > Benjamin Scott writes: > > > I remember a shared login script at UNH which defined various names, > > so you could do things like: > > > > for H in $DWARVES ; do ... ; done > > > > Makes sense to put something like that in /etc/profile or whatever, > > if you're doing to use the fancy name strategy. :) > > > So you basically need a lookup table of some sort. I have an environment that encodes site, server/workstation, base OS/VM in the name. cat hosts | egrep iw # excludes the servers cat hosts | egrep is[0-9][0-9]a gets just the base OS. I've had places that encode the OS (SunOS, Solaris, Linux, OSF, HP-UX, Windows, etc) into the name, cpu relative power, server/workstation/compute node, and location all into 8 characters that can be scripted and/or explained to end users. I've also used NIS netgroups in combo with the names but they didn't go as fine grained. It was useful for a 500 system network with 20+ sysadmins in 1995 Today you might use a more sophisticated database. Or just use something like $DWARVES. The naming scheme means a DB isn't needed if you understand the coding. The 500 node site had some areas with <20 system and they used Loony Tunes. Big characters were servers. Foghorn is obviously bigger then Bugs, right? Physically or Stardom? Another had Star Trek names. I could never get around that because I don't care about Star Trek & don't care to learn character relationships. Then they came up with the naming scheme that could be explained to everyone. The coded names scale up pretty well. This is a good idea, and I ended up doing this too. > > The problem that I had was that I frequently had to deal with the > situation of "this particular problem only really efficiently runs on > 1, 4, or 16 nodes in the cluster" or "this problem only really > efficiently runs on 1, 2, 4, 8, or 16" nodes in the cluster"....now, > what nodes were these again, and how do I relate all of the logfiles > that I obtained from the last program run? > > You might have proven my point. Or you have a grid of Solaris sparc, Solaris x86, Linux (debian, red hat), HP-UX. Some code is shell/perl/python & runs everywhere. Some is compiled only for HP-UX. Or RedHat has this tool needed but Debian doesn't. Luckily, few people have to deal with heterogeneous Unix networks with more then 1 of several types nowadays.
_______________________________________________ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/