[Halld-offline] Fwd: LTO8 tape issue update
Mark Ito
marki at jlab.org
Tue Mar 16 20:02:40 EDT 2021
From Bryan, FYI.
-------- Forwarded Message --------
Subject: LTO8 tape issue update
Date: Tue, 16 Mar 2021 09:51:10 -0400
From: Bryan Hess <bhess at jlab.org>
To: Mark Ito <marki at jlab.org>, Nathan Baltzell <baltzell at jlab.org>, Ole
Hansen <ole at jlab.org>, Brad Sawatzky <brads at jlab.org>
CC: Graham Heyes <heyes at jlab.org>, Jie Chen <chen at jlab.org>, Andy
Kowalski <kowalski at jlab.org>
An Update on the LTO8 tape issue:
We've worked with Chris and Ying and they are creating a database
table/web page for checking the status of the tape recovery. Nothing
fancy, but something that you can view to see what files are in the
recovery process. We should have that for you soon.
Some good news:
* Since the New year we wrote 627TB on 54 LTO8 tapes that were "just
in case" duplicates to test LTO8. We have now read those tapes back
with zero errors. This validates the growing consensus that we are
past the write problem and in the recovery phase.
* Data center humidity trend data since Jan 1 shows that we are
staying in the new, tighter specification (>55%). We will be
watching this closely as we head into summer.
* With some database work I am getting better bounds on the time
window when we had the problem, and the drive serial numbers of
interest. To finish finding the tapes that need recovery, we have
stopped writing the "just in case" duplicates and we are dedicating
that tape bandwidth to auditing suspect tapes. As a result, we have
15 more tapes to go to IBM, and I suspect there will be more.
* IBM has refined their recovery tools, so we are going to return some
tapes for a second pass to see if they can recover a greater
fraction of the files.
Thanks for your patience during this ongoing process. Please let me know
if you have additional questions.
--Bryan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20210316/cc27f10b/attachment-0001.html>
More information about the Halld-offline
mailing list