logo
IRC Archive / Freenode / #opensolaris / 2010 / May / 10 / 4
effnorwood
CosmicDJ: dumpadm = OS and is /var/crash; coreadm = software and is /var/cores as far as I can tell so far
nikolam
I wander will lofi then get up and represent device automatically upon reboot
CosmicDJ, bah, can I mount then dataset in such pool, to directory inside /export/home or I need to mount it in newpool/dir only and to link it?
CosmicDJ
nikolam: no idea, sry
nikolam
ok , thankt
thanks
effnorwood
CosmicDJ: thanks - this is exactly what i wanted
betchou
rm -rf *
nikolam
betchou, not nice.
but we have snapshots, so.. ;)
betchou
oups
nikolam: that'is in a build directory :P
nikolam
within a zone under vbox :)
betchou
ls
nikolam
What algorithm you think one should use for lofiadm encryption and to be fast? (there is zfs compression behind it)
or do you think it is also OK to turn On Zfs encryption on dataset inside lofi file that itself serides on compressed zfs dataset :)
betchou
(Action) is doing bullsh*t with his keyboard
nikolam
But I think it would be just a fine thing to do.
Compress first with Zfas, encryprt with Lofi and then cud down image size with zfs encryption again
I wonder if Lofi device can use more space dinamically and not reside only on one file
CosmicDJ
nikolam: the compression ratio for encrypted stuff should be ~0 %
so you'll get nothing but wasted cpu cycles
madwizard
And a geekfactor +1
:)
nikolam
CosmicDJ, but I will get shrinked lofi image :)
so I use that much space, how much data is inside lofi
:)
If I get lofi image to be dynamically sized depending on data usage, I would not need to use second zfs compression
CosmicDJ
nikolam: It's still not clear to me what you want to do exactly
nikolam
CosmicDJ, to mount my documents to be encrypted on nbook :) simple :)
(Action) currently chooses lofiadm crypto algorithm ..
CosmicDJ
nikolam: http://blogs.sun.com/darren/entry/encrypting_zfs_pools_using_lofi
nikolam
mm nice :) 10x CosmicDJ
wow, this way I have that much data, how much there is space on zfs rpool, very nice :)
no I war wrong, that way he get space out of usable size from disk..
CIA-33
Alexandr Nedvedi*ky <Alexandr.Nedvedi*ky@Sun.COM>: 6912962 Need to compute chksum for packet duped on loopback interface, 6929403 IPF should discard packet silently on OOW event
mui
urr
nachox
congrats on the new job stallion
stallion
nachox: ty
jbk
morning
gdamore
stallion: check your inbox (gmail)
stallion
gdamore: checking
lattera
I'm really considering downloading ON myself and compiling it
lkocman
lattera, do it
stallion
lattera: its a good exercise
much easier now than it used to be ;)
lattera
each day I grow more and more tired of waiting for 2010.1H
asyd
any one use an opensource job scheduler?
glance
torque
lkocman
asyd, I used maestro ... but I don't think that it's opensource
lattera
like, I love opensolaris and its unique features it brings, but this waiting game is just killing me, heh
stallion
lattera: there are reasons for the delay
asyd
tomww:
lattera
ya, I know
asyd
oups
stallion
assuming you want a release that actually works, waiting would be advised.
lattera
of course
asyd
glance: it was for me?
seem yes
lattera
I'm not complaining, I'm just running out of patience... there's some cool ZFS features coming up I'd like to use
glance
we are currently using torque/maui but somewere in the future we are probably switching to slurm
gtirloni
lattera: if you're not using it in production, you can use b134.. closest thing to 2010.whatever (without bug fixes and any backports that Oracle might decide to add)
lattera
ya, I'm using b134 right now, even in production
papertigers
how often does zpool scrub find errors for you guys?
asyd
glance: is the maui also opensource? (or at least free)
lattera
I've never had a disk error, but maybe that's because the disks are new
TomJ
Solaris 10 question, if you'll oblige me: I'm trying to replicate a system, FLAR-stlye, using ZFS send/receive. basic principle: zfs send -R rpool/ROOT/s10_u7_ga, then do a fresh install on a new system, drop to shell, do zfs receive, and then zpool set bootfs=rpool/ROOT/s10_u7_ga. I have done this many times before and it worked
papertigers
lattera: i have had it find errors on my raidz twice now and they are brand new disks
TomJ
but now I am trying it on some IBM servers, and it won't obot - after grub, it just blan screens. failsafe mode says it can't find /boot/x86.miniroot-safe
lattera
interesting
TomJ
the new dataset i created certainly has a /boot/x86.miniroot-safe, so I am not sure why grub would say it can't find it. any thoughts? one difference this time is that I did the new install using U8 before restorting a data set that was U7+Nov09 Patch set.
glance
asyd: http://www.clusterresources.com/products/maui-cluster-scheduler.php
TomJ
when booting in failsafe grub, it reports: "find root (pool,_root,0,a) / filesystem type is zfs, partition type 0xbf / kernel /boot/multiboot -s / module /boot/adm64/x86.miniroot.safe / error:15 ile not found'
so it found and read the FS ok, or so it seems
lattera
seems like it's not looking for /boot/x86.miniroot-safe
TomJ
oh
that's a startingly relevant point
and I dont have a /boot/amd64 on the u7 system i copied. hmm
thanks lattera
stallion
(Action) sighs.
nothing like a nightly on a local vmware guest
jbk
stallion: does it inspire you to 'mv nightly weekly'
=]
stallion
I usually run nightly to get a fresh onbld and then do incrementals after that :P
s/onbld/on build/
CIA-33
Mark Phalan <Mark.Phalan@Sun.COM>: 6940505 MT issues in pkinit
Mark Phalan <Mark.Phalan@Sun.COM>: 6936957 Bad free in gssint_convert_name_to_union_name
Mark Phalan <Mark.Phalan@Sun.COM>: 6932987 memory corruption in kadm5_is_master
Mark Phalan <Mark.Phalan@Sun.COM>: 6941162 krb5_encrypt_helper returns success and an apparently valid buffer on malloc failure
Mark Phalan <Mark.Phalan@Sun.COM>: 6941173 logic error in krb5_build_principal_ext
Mark Phalan <Mark.Phalan@Sun.COM>: 6941180 krb5_gss_accept_sec_context doesn't check return value of krb5_auth_con_getauthenticator()
Mark Phalan <Mark.Phalan@Sun.COM>: 6941191 krb5_send_tgs doesn't do all the error checking it could
Mark Phalan <Mark.Phalan@Sun.COM>: 6941201 double-frees on error path in krb5_os_hostaddr()
Mark Phalan <Mark.Phalan@Sun.COM>: 6941204 krb5_fcc_read_addrs() returns garbage on failure
jbit
epic commit
jbk
=]
papertigers
are there any programs to monitor S.M.A.R.T. from the hard drives?
lincz
Do anybody knows, when b134a will exit from the QA
lattera
no smrt? grrrr
alanc
lincz: no
nachox
alanc, i'm surprised you're not tired of answering to all the fud in osol-discuss
tomww
SFEsmartmontools.spec
alanc
dammit
nachox: it was nicer than "Oracle is fully supporting the Solaris modernization project - just not necessarily YOUR vision of it."
now I've lost whatever the rest of topic used to be
lattera
maybe: We only know 2010.1H should be released in the first half of 2010
so I'm looking at http://hub.opensolaris.org/bin/view/downloads/on
since I'm on b134, I should grab b136's code?
gtirloni
lattera: i'm not sure pkg.opensolaris.org/dev has been updated to anything greater than b134
lattera
right, IPS is still at b134
gtirloni
and even less sure if you can run that with a >b134 onnv
lattera
hrmm
papertigers
what is your guys opinions on nexenta vs opensolaris
stallion
hah
papertigers
stallion: not as good?
stallion
papertigers: oh sorry, no, my comment was about something else
papertigers
ahh okay
stallion
alanc: I did a search for "opensolaris icon" and came across: http://www.alobbs.com/albums/opensolaris_dev07_mugshots/P1090130.jpg
apparently good thinks you are an icon :P
papertigers
do you have an opinion on nexenta though, trying to decide on an os for my storage server
stallion
honestly I don't have much experience with it. they've done quite a bit with storage (which is a good thing)
papertigers
do you run opensolaris as a storage solution
Triskelios
yeah, most people here aren't terribly familiar with nexenta... seems reasonable if you're not already famiiar with osol, as long as you're aware of the limitations in their free vs commercial releases
papertigers
Triskelios: I am just trying to pick the best OS for my setup, it seems the only way to get updates with osol though is to use the dev channel
venture37
trying to boot my poweredge t105 from the b134 iso I get the following error then the system reboots http://img175.imageshack.us/img175/5278/img0099bt.jpg
taemun
poke in the dark, try turning off IOMMU?
Triskelios
papertigers: if you want updates between production releases without buying support, yes
venture37
taemun: how do I dont that?
s/dont/do
taemun
hah nice typo
I'm not sure in the Dell BIOS
but I'll wager theres a button for it
maybe called intel vt-d
(intel vt-x is a different thing)
venture37
t105 is amd opteron based
taemun
my bad
IOMMU is the amd name for it
Triskelios
IOMMU is also the technical non-marketing name for the feature
taemun
^^
venture37
I have high prescission event timer & demand based power management
taemun
tbh I wouldn't be surprised if it was in the bios as VT-d anyway
venture37
not a lot else that's relevant
taemun
umm
Triskelios
venture37: there's no actual error in the screenshot, but presumably the system is about to start the X server at that point
venture37: can you boot with -vk?
taemun
I'll go away, there are folks who know something around :P
venture37
http://img408.imageshack.us/img408/5699/img0100z.jpg
CIA-33
batschul <Frank.Batschulat@Sun.COM>: 6382683 lofi is confused about sync/async I/O
derekv
hmm, so, to fork out a couple hundred bucks more for an enclosure with hot swap bays or spend the money on a couple more hot spares instead..
venture37
Triskelios: http://img408.imageshack.us/img408/5699/img0100z.jpg
taemun
derekv: depends on priorities, surely
having hot spares doesn't really help if you can't power down to swap over the broken hard drive
well .. it kinda does
venture37
Triskelios: I'm at the kmdb prompt now
booted up with -vk
derekv
Since this machine is not gonig to run critical apps, its mostly going to be a ternary backup and archival destination, so 100% availability isn't a huge priority, but I do think about the fact that I am not gonig to be onsite a lot of the time in the future
I'm just thinking out loud.
TomJ
I put a bad disk label over a disk with prtvtoc | fmthard - I copied a 135GB disk label on to a 300GB drive. now I can't get rid of it and get it back to the right label - I tried fdisk -B, I tried deleting the fdisk partition and recreating in format. but when I do the partition list, it still shows s2 (backup) as being 135GB
Triskelios
venture37: https://defect.opensolaris.org/bz/show_bug.cgi?id=10874
alanc
stallion: lol
stallion
alanc: thought you might get a kick out of that ;)
Triskelios
venture37: appears to actually be a be a broadcom driver issue, there is another workaround at http://webcache.googleusercontent.com/search?q=cache:0qZUb_x9UooJ:bugs.opensolaris.org/bugdatabase/view_bug.do%3Fbug_id%3D6852825+6852825 (b.o.o is down right now)
venture37
Triskelios: trying it now
YAY
:)
Triskelios
venture37: you have some familiar-looking relics in the right of the photo ;P
venture37
Triskelios: http://img707.imageshack.us/img707/2537/img0102lm.jpg
one of the ultra 1s has a dead nvram & the other either needs the ram reseating or dead motherboard, either way, I will be able to bring one back to life using parts from the other
derekv
how often do you think i'd need to pull out the server if the drives are all internal, sata consumer grade disks, ... with 10 disks, say afr of around 4%, 40% of years, thats less than once a year.
I bet afr is higher than that.
Stric
I have a sata storage (8disk) that has worked for 5+ years of pounding without any issues.. so it all depends..
TomJ
I have 16 x Maxtor 400GB drives in an array which I got second hand, and must have been at least 3 hours before I got them - 3 years ago - and I've had only 1 drive failure in all that time. and before that I thougth maxtor drives were the sh*ttest of the sh*ttest.
s/hours/years/
admittedly I use them very lightly now - home media - but they're probably too hot
derekv
hmm, so maybe.
Triskelios
venture37: apparently you can replace the battery in the old NVRAM modules with a regular CR2032
venture37
Triskelios: yes, too much hassle
vraa
zfs is a mixture between heaven and peanut butter
it's so amazing, the snapshot feature is so perfect, combined with dedupe, combined with compression, it's just all so sweet
i have files stored on a windows pc, and the osol box running zfs just rsyncs --delete every few minutes from the winpc to the osol box and it auto makes the 15 minute snapshots
Triskelios
dedup is far from perfect right now
Stric
and if you're doing it wrong, it gets sticky
vraa
i set dedup=verify
alanc
feeding it to a dog isn't as amusing to watch as pure peanut butter though
effnorwood
vraa - just don't try to zfs destroy on a dedup file system - you will be very sad
venture37
Triskelios: Thanks for your help
I gtg
RoyK
Triskelios: dedup isn't very close to perfect now
vraa
why would i want to destroy my data lol? :) btw, i am doing "pfexec zfs list" and it used to give me the dedupratio but now i can't find it. i also did "pfexec zfs get dedupratio" and it pretends like nothing exists
RoyK
vraa: zpool list
dedup is done in the pool, not in the dataset
vraa
ah thanks, lol so many times i confuse zfs and zpool, my dedup ratio is 3.77x
wrapster
im running into locale issues again.. and this time please take a loo at it.. it explains evertyhing... im unable to get thing working with ssh ... no matter what i try... If i export LANG then everything is fine.. By default after i change the init file and reboot.. nothing seems to occur...http://pastie.org/954070
pls help..
edude03
Hey everyone
Can anyone recommend a (solaris compatible) relatively affordable sata card
stallion
edude03: AHCI chips are the best supported, but there aren't many available as a card
baring that, it depends on what you want to use the controller for
if you don't need ATAPI, the mv88sx chips are fine
edude03
ZFS pool