Date: Fri, 02 Nov 2012 20:00:27 +0100 From: Bas Smeelen <b.smeelen@ose.nl> To: Mateusz Guzik <mjguzik@gmail.com> Cc: freebsd-stable@freebsd.org Subject: Re: [patch] Re: SU+J on 9.1-RC2 ISO Message-ID: <5094184B.6070100@ose.nl> In-Reply-To: <20121102184131.GB22755@dft-labs.eu> References: <5093F934.7050306@ose.nl> <5093FD3D.3080201@ateamsystems.com> <1351876381.2657.1.camel@mjakubik.localdomain> <50940276.5030306@ateamsystems.com> <50940C20.3090409@ose.nl> <50940E40.3090709@ose.nl> <5094112C.2070102@ose.nl> <20121102184131.GB22755@dft-labs.eu>
next in thread | previous in thread | raw e-mail | index | archive | help
On 11/02/2012 07=3A41 PM=2C Mateusz Guzik wrote=3A =3E On Fri=2C Nov 02=2C 2012 at 07=3A30=3A04PM +0100=2C Bas Smeelen wrote= =3A =3E=3E On 11/02/2012 07=3A17 PM=2C Bas Smeelen wrote=3A =3E=3E=3E On 11/02/2012 07=3A08 PM=2C Bas Smeelen wrote=3A =3E=3E=3E=3E On 11/02/2012 06=3A27 PM=2C Adam Strohl wrote=3A =3E=3E=3E=3E=3E On 11/3/2012 0=3A13=2C Mike Jakubik wrote=3A =3E=3E=3E=3E=3E=3E You can disable SU+J after installing=2C though it would= be nice if the =3E=3E=3E=3E=3E=3E installer gave you a choice=2E =3E=3E=3E=3E=3E This assumes that you know about this flaw=2C which most pe= ople do not=2E =3E=3E=3E=3E=3E =3E=3E=3E=3E=3E I didn=27t until I discovered it by panic-ing a perfectly f= ine running =3E=3E=3E=3E=3E server=2E Getting burned by a known bug like this shouldn= =27t be =22SOP=22 =3E=3E=3E=3E=3E for users of FreeBSD=2E =3E=3E=3E=3E=3E =3E=3E=3E=3E=3E If anything it should be turned off by default=2C and peopl= e can turn it =3E=3E=3E=3E=3E on if they want given the landmine it plants=2E If they kn= ow how to =3E=3E=3E=3E=3E turn it on they=27re much more likely to be aware of the is= sue=2E =3E=3E=3E=3E=3E =3E=3E=3E=3E=3E =3E=3E=3E=3E=3E =3E=3E To sum it up =3E=3E SU+J should be turned off by default because of =3E=3E 1=2E It does not work with dumping a live system e=2Eg=2E snapshot= =3E=3E 2=2E it is not recommended for SSD installs =3E=3E 3=2E =22Smart=22 admins can turn it on if they want =3E=3E =3E=3E root=40sys=3A/usr/src/usr=2Esbin/bsdinstall/partedit =23 diff -u gpa= rt=5Fops=2Ec =3E=3E gpart=5Fops=2Ecnew =3E=3E --- gpart=5Fops=2Ec 2012-08-06 01=3A54=3A33=2E000000000 +0200 =3E=3E +++ gpart=5Fops=2Ecnew 2012-11-02 19=3A07=3A45=2E000000000 +0100= =3E=3E =40=40 -90=2C8 +90=2C8 =40=40 =3E=3E =7B=22SU=22=2C =22Softupdates=22=2C =3E=3E =22Enable softupdates =28default=29=22=2C 1 =7D= =2C =3E=3E =7B=22SUJ=22=2C =22Softupdates journaling=22=2C =3E=3E - =22Enable file system journaling =28default - =22= =3E=3E - =22turn off for SSDs=29=22=2C 1 =7D=2C =3E=3E + =22Disable file system journaling =28default - =22= =3E=3E + =22turn on for adventurish admins=29=22=2C 0 =7D=2C= =3E=3E =7B=22TRIM=22=2C =22Enable SSD TRIM support=22=2C =3E=3E =22Enable TRIM support=2C useful on solid-state d= rives=22=2C =3E=3E 0 =7D=2C =3E=3E =3E=3E Please comment=2C then I can file a PR or not =3E As was noted in my another mail=2C the kernel will no longer crash when= an =3E attempt to take a snapshot is made=2E Also AFAIR SUJ can be disabled =3E later=2E =3E =3E Given that I prefer the following=3A =3E =3E diff --git a/usr=2Esbin/bsdinstall/partedit/gpart=5Fops=2Ec b/usr=2Esbi= n/bsdinstall/partedit/gpart=5Fops=2Ec =3E index 479365a=2E=2E80296c2 100644 =3E --- a/usr=2Esbin/bsdinstall/partedit/gpart=5Fops=2Ec =3E +++ b/usr=2Esbin/bsdinstall/partedit/gpart=5Fops=2Ec =3E =40=40 -91=2C7 +91=2C7 =40=40 newfs=5Fcommand=28const char *fstype=2C c= har *command=2C int use=5Fdefault=29 =3E =09=09=09 =22Enable softupdates =28default=29=22=2C 1 =7D=2C =3E =09=09=09=7B=22SUJ=22=2C =22Softupdates journaling=22=2C =3E =09=09=09 =22Enable file system journaling =28default - =22 =3E -=09=09=09 =22turn off for SSDs=29=22=2C 1 =7D=2C =3E +=09=09=09 =22turn off for SSDs or if you use snapshots=29=22=2C 1= =7D=2C =3E =09=09=09=7B=22TRIM=22=2C =22Enable SSD TRIM support=22=2C =3E =09=09=09 =22Enable TRIM support=2C useful on solid-state drives= =22=2C =3E =09=09=09 0 =7D=2C =3E =3E http=3A//people=2Efreebsd=2Eorg/=7Emjg/patches/suj-snapshot-comment=2Ed= iff =3E Hi Mateusz I can see some benefits of having journaled soft updates and I=20 understand your preference=2E Though the last 10 years I have not had the inconvenience of having to=20= deal with long fsck=27 s or bgfsck=27 s on servers or workstation installs= =2C=20 so I think this should not be default on new installs=2E Since there are more SSD installs nowadays=2C it might be a good option to= =20 turn soft updates journaling off by default=2E Can you explain the benefits of having journaled soft updates turned on=3F= There are also a lot of back-up service providers who use the snapshot=20= functionality to be able to host a lot of back-ups=2C though they might be= =20 using zfs for this now I would guess=2E Kind regards Bas This e-mail message=2C including any attachment=28s=29=2C is intended solel= y for the addressee or addressees=2E Any views or opinions presented herein= are solely those of the author and do not necessarily represent those of O= SE=2E If you are not the intended recipient of this communication please return t= his e-mail message and the attachment=28s=29 to the sender and delete and d= estroy all copies=2E
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5094184B.6070100>