leaf services down for a few hours

Matthew Dillon dillon at apollo.backplane.com
Fri May 15 10:57:48 PDT 2015


Leaf services, including the web site, will be down for a few hours while
a bad disk used by /build is replaced.  /build is not a backed-up partition
(which we warn about), and theoretically only contains stuff people don't
care about much or can restore.  But I think its worth a few hours of down
time to try to keep the data intact so I'm going to see if cpdup can grind
through the read errors.  If not, I'll scrap it so leaf can be brought up to
full services again and do an off-line restore.

The drive is only ~2 years old but I made the mistake of not turning off APM
so it has 1.5M load cycles on the heads, which is probably why it's going
south.

By next year all of the blade server's disks except for the backup drives
will be changed over to SSDs.  Going to baby it until then.

Note that people's GIT repos and ~/public_html directories are on /home,
which is a different (and backed-up) disk.

					-Matt
					Matthew Dillon 
					<dillon at backplane.com>


More information about the Users mailing list