[16] in Sysdev statistics

home help back first fref pref prev next nref lref last post

Locker Hours statistics

daemon@ATHENA.MIT.EDU (lwvanels@Athena.MIT.EDU)
Tue Jan 7 11:00:28 1992

From: lwvanels@Athena.MIT.EDU
Date: Tue, 7 Jan 92 11:00:22 -0500
To: sysd-stats-mtg@brennin.LOCAL
Reply-To: lwvanels@mit.edu

These statistics are for about the last three days in November and all of
December:

------------------------------------------------------------------------------
                        File Server Availablility Statistics

      Server       Up   Down    Net     % up  Lockers      LHU     % LHU

   aphrodite   824.00   3.00   0.00   99.64%     2445     7160     0.37%
       atlas   823.33   3.67   0.00   99.56%     2782    10262     0.44%
       cyrus   824.83   2.17   0.00   99.74%     1673     3643     0.26%
      hecate   823.17   3.83   0.00   99.54%     1435     5497     0.46%
    odysseus   826.25   0.75   0.00   99.91%      851      670     0.09%
       rodan   826.75   0.25   0.00   99.97%       66       16     0.04%
       talos   816.83  10.17   0.00   98.77%     1269    18628     1.58%
      themis   823.92   3.08   0.00   99.63%     1731     5375     0.37%

Total Locker Unvailability: 0.50%
------------------------------------------------------------------------------

Server		-	self-explanatory
up		-	number of hours the server was nfs-pingable
down		-	number of hours the server was not nfs-pingable, but
			reachable through the network from the NOC
net		-	number of hours the server was unreachable from
			the NOC
% up		-	percentage of the time server was in the "up" state
lockers		-	number of lockers on the server at the last
			sampling time (in this case, December 31)
LHU		-	locker hours unavailable - this is basically just
			down * lockers, taking into account that the number
			of lockers on a server changes over time
% LHU		-	this is 100*LHU/TLH (Total Locker Hours on that server)

Total locker unavailability is the total locker unavailability across
all servers.

How the numbers are taken:

	NOC pings the selected NFS servers once every five minutes
	and records the results. When the results are compiled,
	each sample represents its five minute interval.

For these samplings, unfortunately, the program wasn't configured quite
correctly on the NOC and the "net" statistic was unavailable. So anything
that would have been "net" shows up in "down" for this month.

home help back first fref pref prev next nref lref last post