[1595] in SIPB-AFS-requests
Cutover plans
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Tue Nov 15 17:45:00 1994
From: ghudson@MIT.EDU
Date: Tue, 15 Nov 94 17:42:06 -0500
To: sipb-afsreq@MIT.EDU
Cc: yandros@MIT.EDU
Due to a change in policy in purchasing (they now want to do
competitive bids before making purchases), the delivery of the 4GB
drive has been delayed. Purchasing claims it will only take them a
day or two to make the order, but such things tend to get delayed.
If they do place the order this week, we should stay with our current
plans. If they don't place the order, then we can do the AFS cutover
in the next couple of weeks night without using the 4GB AFS disk.
rosebud currently has 2.42GB of data on it (where 1GB == 1000000KB).
Thus, it would be nice to have 3GB of swap disk: one from beacon's old
disk, one from ops-tmp-w20's hot spare, and one we borrow from ops.
However, the one we borrow from ops may not be available (the Athena
cell is running out of space right now), in which case we may need to
be clever.
If we tunefs them to eliminate most of the 10% margin, two external
1GB disks can account for a full 2GB of storage, instead of the 1.85GB
they would normally account for. This leaves 0.42GB of data. We can
fit 0.1GB on ops-tmp-w20's internal disk, leaving 0.32GB. ronald-ann
currently has 0.37GB free, and we may be able to further reduce space
needs by temporarily eliminating the replication sites on project.sipb
and other volumes.
Also, there is probably some data which can go away. For instance,
machine.lola-granola currently has 136MB of data which is probably no
longer needed.
Nevertheless, it would be tight and would require a lot of attention
to detail; I would much rather have 3GB of swap space, or, better yet,
the 4GB disk.
