Someone posted about CERN having a bad network card which injected faulty bits into the data stream. And ZFS detected it, because of end-to-end checksum. Does anyone has more information on this? -- This message posted from opensolaris.org
Hi,> Someone posted about CERN having a bad network card which injected faulty bits into the data stream.I''ve regularly heard people mention data corruption in network layer here - and other physics sites like FNAL and SLAC - but I don''t have details of any specific incident. We do see our files get corrupted in transit every once in a while though. May be it was about this paper which details some of the findings? http://indico.cern.ch/getFile.py/access?contribId=3&sessionId=0&resId=1&materialId=paper&confId=13797 I can''t really speak to where ZFS would be in use in CERN''s many infrastructures. I believe the big experiment data pools are all linux disk servers. Regards, Lassi
On Wed, Aug 11, 2010 at 4:05 AM, Orvar Korvar < knatte_fnatte_tjatte at yahoo.com> wrote:> Someone posted about CERN having a bad network card which injected faulty > bits into the data stream. And ZFS detected it, because of end-to-end > checksum. Does anyone has more information on this? > -- > This message posted from opensolaris.org > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss >I''m not sure abotu thsi particular case but i can tell you that i had a raid card start to fail and corrupt some data and ZFS solved it, and due to frequent scrubs i was able to see that corruption was happening. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20100811/f39850f9/attachment.html>
On Aug 11, 2010, at 04:05, Orvar Korvar wrote:> Someone posted about CERN having a bad network card which injected > faulty bits into the data stream. And ZFS detected it, because of > end-to-end checksum. Does anyone has more information on this?CERN generally uses Linux AFAICT: http://linux.web.cern.ch/linux/scientific.shtml