|
on solaris, zfs was a software marvel that threatened the business models of a dozen NAS vendors. mature, well-supported, battle-tested in the field. on linux, zfs is a fork of a long-dead open source project supported by one (1) guy at a national research lab
|
#
?
Nov 8, 2017 20:48
|
|
|
|
| # ? Apr 19, 2024 13:28 |
|
|
Notorious b.s.d. posted:on solaris, zfs was a software marvel that threatened the business models of a dozen NAS vendors. mature, well-supported, battle-tested in the field. god bless Brian Behlendorf
|
|
#
?
Nov 8, 2017 20:52
|
|
|
Notorious b.s.d. posted:on linux, zfs is a fork of a long-dead open source project supported by one (1) guy at a national research lab much like a lot of other critical, widely used components such as bash and the timezone database zfs is still being actively developed as open-zfs. in fact they are finally working on adding raid-z expansion
|
|
#
?
Nov 8, 2017 20:55
|
|
|
The_Franz posted:much like a lot of other critical, widely used components such as bash and the timezone database haven't checked, but I'm pretty sure btrfs gets a lot more developer action than zfs.
|
|
#
?
Nov 8, 2017 22:41
|
|
|
Notorious b.s.d. posted:on a dead operating system, yes zfs is mature and well supported on freebsd. freebsd is good and not dead. Tankakern posted:haven't checked, but I'm pretty sure btrfs gets a lot more developer action than zfs. maybe so, but zfs raid actually works and doesn't eat your data.
|
|
#
?
Nov 8, 2017 22:56
|
|
|
Tankakern posted:haven't checked, but I'm pretty sure btrfs gets a lot more developer action than zfs. they're all crapping in it, as far as i can tell
|
|
#
?
Nov 8, 2017 22:58
|
|
|
Condiv posted:is butterfs dead or can I use it for a shared storage system on my network? 💀 for fun try some sort of distributed filesystem instead
|
|
#
?
Nov 8, 2017 23:02
|
|
|
why would i need btrfs raid or zfs raid if i have an actual hw raid controller?
|
|
#
?
Nov 9, 2017 01:08
|
|
|
filesystem level snapshots, cross-machine or FS incremental snapshot replication, data and metadata chrcksumming in a smart way, rebuilding data instead of a block device, saving time if the FS on the disks isn't very full the filesystem level redundancy and checksumming lets the filesystem recover from minor corruption the way hardware raid can't: if you have a RAID 1 and the two copies of your data differ, with hardware raid it's a crapshoot, with zfs or btrfs you use the copy with the correct checksum to fix the corrupt copy
|
|
#
?
Nov 9, 2017 01:17
|
|
|
Lysidas posted:filesystem level snapshots, cross-machine or FS incremental snapshot replication, data and metadata chrcksumming in a smart way, rebuilding data instead of a block device, saving time if the FS on the disks isn't very full hmm, cool. looking at the info on btrfs quotas though is discouraging. it seems like subvolumes and quotas can interact to result in lost disk space, which is p hosed up
|
|
#
?
Nov 9, 2017 01:36
|
|
|
i'm really lost on btrfs. in one place i read "the issues with x are really bad and it'll eat your data", in another i read "x works fine don't worry about it" does SLES use any special patches on btrfs or is it just bog standard btrfs i could get in centos 7 or fedora server edition or something? i'm thinking about trying to make this a RAID 10 system and i hope it's worth it 
|
|
#
?
Nov 9, 2017 02:30
|
|
|
BuTteRy loving Smooth
|
|
#
?
Nov 9, 2017 03:24
|
|
|
The_Franz posted:just use zfs on FreeBSD
|
|
#
?
Nov 9, 2017 03:34
|
|
|
SamDabbers posted:zfs is mature and well supported on freebsd. freebsd is good and not dead. exactly—use ZFS on FreeBSD for storage, use that storage on other systems via NFS or whatever
|
|
#
?
Nov 9, 2017 03:35
|
|
|
btrfs has inaccurate free space calculations and forces you to consider metadata space when fixing filesystem issues and does not want you to use all of your free space zfs is kinda confusing when I used it for freenas xfs is straightforward and just works
|
|
#
?
Nov 9, 2017 03:40
|
|
|
Lysidas posted:the filesystem level redundancy and checksumming lets the filesystem recover from minor corruption the way hardware raid can't: if you have a RAID 1 and the two copies of your data differ, with hardware raid it's a crapshoot, with zfs or btrfs you use the copy with the correct checksum to fix the corrupt copy Also known as the RAID write hole, or in House's words: everyone lies. Hardware and RAID driver developers have been breaking the commit contract. The original ZFS blog was quite enlightening on topic.
|
|
#
?
Nov 9, 2017 04:13
|
|
|
btrfs and zfs stir up my curmudgeonly feelings but idk maybe they're good i kinda feel like you shouldn't have 16+ tb filesystems and have that filesystem be the highest level of abstraction you work with tho
|
|
#
?
Nov 9, 2017 04:17
|
|
|
it's called btrfs because it's not good but it's getting btr
|
|
#
?
Nov 9, 2017 04:46
|
|
|
Cocoa Crispies posted:it's called btrfs because it's not good but it's getting btr i worked with zfs on solaris, and it was cool and good. super easy to work with. first time i did a disk swap i was confused at the end because it couldn't possibly have been that easy and smooth dunno about zfs on other platforms tho
|
|
#
?
Nov 9, 2017 05:07
|
|
|
RFC2324 posted:i worked with zfs on solaris, and it was cool and good. super easy to work with. first time i did a disk swap i was confused at the end because it couldn't possibly have been that easy and smooth yeah former employer's product allegedly was best on zfs and smartos but now i mostly don't care about filesystems because the exciting poo poo happens in process memory
|
|
#
?
Nov 9, 2017 05:10
|
|
|
RFC2324 posted:i worked with zfs on solaris, and it was cool and good. super easy to work with. first time i did a disk swap i was confused at the end because it couldn't possibly have been that easy and smooth thats basically it. my main problem with zfs now is for some reason dkms doesnt rebuild it automatically anymore
|
|
#
?
Nov 9, 2017 06:25
|
|
|
Sapozhnik posted:btrfs and zfs stir up my curmudgeonly feelings but idk maybe they're good it works like lvm, where for home use you might as well use all your available storage
|
|
#
?
Nov 9, 2017 06:26
|
|
|
SamDabbers posted:zfs is mature and well supported on freebsd. freebsd is good and not dead. freebsd zfs is derived from the same ancient lovely code dump as the Linux version, circa 2010. i am not willing to bet my data or my business on the "maturity" of a fork from an ancient beta of solaris, whether that forked driver is hacked into FreeBSD or Linux doesn't matter
|
|
#
?
Nov 9, 2017 07:19
|
|
|
Condiv posted:why would i need btrfs raid or zfs raid if i have an actual hw raid controller? At this point using hardware raid 5 or 6 with large (multiple terabytes) disks is downright dangerous, because the chances of a disk error increase with size and the time for recovery does as well. And because spinners haven't really gotten much faster recovery takes ages. So if (when) a disk fails and the array needs to be rebuilt it will take days to complete and the chances of another disk in the array developing an issue while rebuilding is pretty high.
|
|
#
?
Nov 9, 2017 07:32
|
|
|
spankmeister posted:At this point using hardware raid 5 or 6 with large (multiple terabytes) disks is downright dangerous, because the chances of a disk error increase with size and the time for recovery does as well. And because spinners haven't really gotten much faster recovery takes ages. So if (when) a disk fails and the array needs to be rebuilt it will take days to complete and the chances of another disk in the array developing an issue while rebuilding is pretty high. this happens all the time with our old EVA SAN. rebuild typically takes one or two other disks with it and takes days, even with 15k disks. would love a ssd san but our ceo is pretty committed to buying used poo poo from 2015
|
|
#
?
Nov 9, 2017 07:35
|
|
|
Sapozhnik posted:btrfs and zfs stir up my curmudgeonly feelings but idk maybe they're good again poo poo like lizardFS is pretty dang cool
|
|
#
?
Nov 9, 2017 11:03
|
|
|
raid with btrfs is cool+good because you get free corruption detection and automatic repair at the file level. you don't get that with hw raid. don't use lvm if your just jbodding stuff, make one gpt partition (to be sure that stuff is aligned to 4k) and just use btrfs freespace calc is much better now too, all that stuff people here keeps nagging about btrfs is fixed problems. and doing backups on btrfs is so good that i can never return to another filesystem now
|
|
#
?
Nov 9, 2017 12:52
|
|
|
once bitten
|
|
#
?
Nov 9, 2017 17:37
|
|
|
year of linux on the ppc laptop: https://www.powerpc-notebook.org/en/ i couldn't find anything on the website to explain *why* a ppc laptop is something that needs to happen beyond "ppc is a cool architecture." any insights?
|
|
#
?
Nov 12, 2017 19:29
|
|
|
Best Bi Geek Squid posted:year of linux on the ppc laptop: https://www.powerpc-notebook.org/en/ modern x86 is closed as gently caress. the newest laptop you can boot with purely open source software + firmware is from circa 2006. if you are shopping for an arch to build a meaningfully "open" laptop on top of, your options are mips, power, and arm. actually-existing mips and arm cores that you can really licese are slow as dirt. so that leaves power as the last man standing.
|
|
#
?
Nov 12, 2017 19:43
|
|
|
a company tried to market a power8 workstation not too long ago and failed miserably, mainly because they were asking $3000+ for a machine with limited software support. they are trying again with a power9 based machine, starting at the low, low price of only $4100. are the current ppc perf/watt numbers even suitable for a laptop? one of the main reasons why apple dumped them years ago was because they were falling way behind intel in that department.
|
|
#
?
Nov 12, 2017 19:52
|
|
|
The_Franz posted:a company tried to market a power8 workstation not too long ago and failed miserably, mainly because they were asking $3000+ for a machine with limited software support. they are trying again with a power9 based machine, starting at the low, low price of only $4100. i think they failed less because of the price, and more because they are a fly-by-night with no product they were asking people to lay down $4k pre-orders to fund the development of the thing ostensibly being offered for sale The_Franz posted:are the current ppc perf/watt numbers even suitable for a laptop? one of the main reasons why apple dumped them years ago was because they were falling way behind intel in that department. the big ibm power chips certainly aren't suitable freescale might have something, idk
|
|
#
?
Nov 12, 2017 19:54
|
|
|
Notorious b.s.d. posted:i think they failed less because of the price, and more because they are a fly-by-night with no product thats ridiculous who would ever do something like that oh. right.
|
|
#
?
Nov 12, 2017 19:57
|
|
|
https://twitter.com/ppcinstructions
|
|
#
?
Nov 12, 2017 20:17
|
|
|
The competence gap between Intel and literally every single other CPU manufacturer put together really is pretty remarkable. Apart from that one slip-up with NetBurst they really were (and are) a force to be reckoned with. I can't think of any other market sector where one company so completely and utterly dominates its competition.
|
|
#
?
Nov 12, 2017 20:22
|
|
|
"Sure is a nice semiconductor business you've got going on here, very profitable. Be a real shame if it was suddenly threatened with a huge anti-trust investigation, yeah? Does horrible things to the share price. Say, on another topic, we had some ideas about our SIGINT enablement efforts we wanted to run by you."
|
|
#
?
Nov 12, 2017 20:25
|
|
|
I once got an arm laptop (probably more accurately called a netbook) from a company called Genesi. It was thinner than a comparable acer netbook, but screen size was about the same, I think. Battery life was amazing. As long as I didn’t have to compile anything or visit complex websites, it was pretty OK. I mainly used it for for typing and taking notes.
|
|
#
?
Nov 12, 2017 20:36
|
|
|
Apple seems to be decently competent at CPU design these days
|
|
#
?
Nov 12, 2017 20:49
|
|
|
amd invented the architecture everyone uses. well cya
|
|
#
?
Nov 12, 2017 20:51
|
|
|
|
| # ? Apr 19, 2024 13:28 |
|
|
Sapozhnik posted:The competence gap between Intel and literally every single other CPU manufacturer put together really is pretty remarkable. Apart from that one slip-up with NetBurst they really were (and are) a force to be reckoned with. I can't think of any other market sector where one company so completely and utterly dominates its competition. Suspicious Dish posted:Apple seems to be decently competent at CPU design these days yeah, in fact they're doing vastly better than intel at cpu cores designed for phone power profiles. intel completely missed out on that market, the stuff they tried to sell was late and kinda half-assed and eventually they gave up when even heroic marketing techniques (essentially giving the chips away) failed to make a dent. in the high power realm intel has had little meaningful competition during the ~10 years of amd wandering in the desert, but apparently amd's newest cpu core is p. deece? idk much about it, haven't had time to read up
|
|
#
?
Nov 12, 2017 21:26
|
|
