Shane - zpool history and snapshots

New About Yours API Help
1.5 KB, Plain text
Hey guys, great show.

Before the questions let me share a "Success from the trenches!!". A few
days ago I had a disk die, not spitting errors, it just stopped existing
as far as the system was concerned. So anyway one disk at a time I
replaced the three disks in my raidz pool and am happily running on new
spinning rust without loosing anything. Sure things were slow for a few
days while resilvering but everything just kept working.

So I got curious and had a look at my pools history, well it was created
in 2013 so it will turn five in a couple of months. As you can probably
guess, the history took a while to scroll by. This got me to thinking
about how much space it used, well at over 250,000 lines dumping it to a
text file takes 30MB but having a lot of repetition it compresses to 1MB
and I guess the pool could do better than that as it would build it from
existing data.

But then I figured, after ten years it would have half a million lines
of history, after twenty over a million, that's a lot of history to get
through if I want to see the last few steps, and the history command
only lists one way. While disk usage may not be an issue, should we
consider replacing our pools at times to prevent the history getting to
an un-managable length?

A while back you talked about work being done on zpool snapshots. Would
this support resilvering? For example remove an old drive and start
resilvering but one of the other drives dies, preventing the resilver
from finishing.
Pasted 4 months, 3 weeks ago — Expires in 222 days