Date: Thu, 5 Jul 2018 10:09:18 -0700 (PDT) From: "Rodney W. Grimes" <freebsd-rwg@pdx.rh.CN85.dnsmgr.net> To: Peter Jeremy <peter@rulingia.com> Cc: Eugene Grosbein <eugen@grosbein.net>, FreeBSD Hackers <freebsd-hackers@freebsd.org>, George Mitchell <george+freebsd@m5p.com> Subject: Re: Confusing smartd messages Message-ID: <201807051709.w65H9IfA048597@pdx.rh.CN85.dnsmgr.net> In-Reply-To: <20180705063352.GA76985@server.rulingia.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 2018-Jul-04 18:05:56 -0700, "Rodney W. Grimes" <freebsd-rwg@pdx.rh.CN85.dnsmgr.net> wrote: > >It would be really nice if we had a way to tell ZFS to > >do the equivelent of: > > dd if=/dev/ada1 of=/dev/ada1 bs=128k conv=noerror,sync > >in some type of lowlevel scrub operation, with proper locking > >which could easily repair most of these Pending Sector errors, > > Agreed. > > >It would of also been nice if the ata standard would of made a way > >to get the LBA of pending sectors so that a very quick rewrite attempt > >could be done to fix them. IIRC this info is avalible, but in a vendor > >specific way. > > As a workaround, > dd if=/dev/ada1 of=/dev/null bs=128k conv=noerro > will report any errors. You can then write zeroes or garbage to those > sectors (optionally followed by a scrub if you overwrote more than the > faulty sectors) Though this is a good test to run be-aware that this only report errors if infact the read fails, and often pending sectors do not fail as they are a transient error. This is also a very long and slow test, I can go to a PC3000 driver recovery based system and have the pending sector list in a mater of a single command and less than 1s of time. I can also some times obtain that info by hooking up a TTL serial cable to the jumpers on the drive and issuing commands to the controller chip. But again, this is a very vendor specific situation and rather intrusive to do. -- Rod Grimes rgrimes@freebsd.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201807051709.w65H9IfA048597>