We had an intense line of thunderstorms come through in the night of
June 10th. It took out power across campus. In our building, log files
indicate the power was out from about 23:00 until 00:50, so nearly 2
hours. Too long for the battery backups to carry everything.
Amanda backups were running when the power dropped. I have no way of
knowing if the tape was writing when the power dropped or was just
sitting in the drive. Anyway, amcleanup did it's job. The serious
problem is with the library. It was improperly powered off, and it now
doesn't seem to sense the tape that was in the drive. I have to sort
that out. In the meantime, mt and mtx aren't happy. The puzzling thing
is that when Amanda ran starting the evening of June 11, it did not fall
back to incrementals only. It got an error on the tape check, but
proceeded to do a mix of fulls and incrementals. When I did amstatus, it
showed current things that were "dump done" and "wait for writing to
tape", along with things from the 10th that were "waiting to flush". I
happen to have plenty of holding disk space, so it's not a problem. But
it's also not what I expected to have happened.
The mtx debug was as follows:
mormyrid:/tmp/amanda/server:amanda$ more chg-zd-mtx.20080611224501.debug
chg-zd-mtx: debug 1 pid 998 ruid 555 euid 555: start at Wed Jun 11
22:45:01 2008
22:45:04 Using config file /usr/local/etc/amanda/daily/changer.conf
22:45:04 Arg info:
$# = 1
$0 = "/usr/local/libexec/chg-zd-mtx"
$1 = "-info"
22:45:04 Running: /usr/local/sbin/mtx status
22:45:04 Exit code: 1
Stderr:
no Data Transfer Element reported
22:45:04 Exit (2) -> <none> no slots available
chg-zd-mtx: pid 1082 finish time Wed Jun 11 22:45:04 2008
As a follow up question to this, is there even a moderately clean way of
telling Amanda to cut its losses and terminate if we have UPS software
running and get the message that a power outage is impending? I know
that terminating ufsdump is more than a little problematic. So there may
be a limit to what Amanda can do. But, if it could at least terminate
taper and either put the library in a static position with mtx or allow
me to run a script that would do that, it might save some trouble.
---------------
Chris Hoogendyk
-
O__ ---- Systems Administrator
c/ /'_ --- Biology & Geology Departments
(*) \(*) -- 140 Morrill Science Center
~~~~~~~~~~ - University of Massachusetts, Amherst
<hoogendyk AT bio.umass DOT edu>
---------------
Erdös 4
|