asciilifeform: so gotta happen in dulap.

asciilifeform: BingoBoingo: if you can swing by the dc today, stuff 2 of the new drives into dulap plox.

asciilifeform: 4.9.76 on dulap, and slightly earlier iirc in rk branch ( will have to maintain at least these 2 forks , i suspect )

asciilifeform: trinque: fwiw i was able to measure mtu ~from~ dulap, using USE="tracepath" emerge iptools and tracepath -4 asciilifeformshouse

asciilifeform: trinque: i just noticed this, on dulap also, apparently some recent year's shitgnomism in gentoo mutilated traceroute

asciilifeform: mircea_popescu: there's some % who actually endrivered various new iron. e.g. dulap kernel built for arm64 won't actually boot on rk, needs the little bit of chinesium ( i have it laid out, fwiw , mostly, specifically for when we pour cement )

asciilifeform: ave1: nope, /usr/include/x86_64-linux-gnu exists in trad gentoo as well (on e.g. dulap)

asciilifeform: BingoBoingo: dun have to be right now, can go in the morning. and when you do, also plz retrieve the 2 retired usb stix from dulap, they've been zero'd

asciilifeform: at one time i had a 'nuke anyffing that behaves like bot' on dulap

asciilifeform: one of the 'dead' sticks seems to read entirely ok in dulap

asciilifeform: BingoBoingo: at your earliest chance plz pull mod6's stick from dulap, re-emplace into his unit, and power up

asciilifeform: mircea_popescu: note that this kernel is tailored to dulap-style box (i.e. amd, and with that particular raid card)

asciilifeform: BingoBoingo: feel free to route all unrouted ips to dulap

asciilifeform: http://btcbase.org/log/2018-08-19#1842911 << BingoBoingo plox to do as we did with prev rk, put the disk in dulap and ping asciilifeform when did. ☝︎

asciilifeform: mircea_popescu: primary dulap has 4 holes, all fulla ssd ; spare dulap has 4 empty holes

asciilifeform: ... in fact, loox like there won't be any moar phuctor output until i get a 1TB disk hooked up ( will have to be on a usb snake, luckily i left BingoBoingo with one -- as all 4 drive bays on dulap are full )

asciilifeform: if this worx for you, go ahead and move your drive over to dulap

asciilifeform: BingoBoingo: i'ma test new kernel today, then tomorrow will ask BingoBoingo to power down rk-c and put disk in dulap to i can snap new master img, then it will go into service ( and can reimage BingoBoingo also ) . ☟︎☟︎

asciilifeform: ( bad enuff that i only even have usb2, not 3, on dulap, and it takes 3hrs )

asciilifeform: ( i can't afford it on dulap, because it also doubles as reimaging station for rockchippen , but s.mg prolly can safely say goodbye to usb3 ) ☟︎

asciilifeform: dulap ~ # zcat /proc/config.gz | grep CONFIG_USB_SERIAL_OPTION

asciilifeform: ( at least dulap kernel -- does not )

asciilifeform: ( fwiw i've dd's from dulap's FGs previously, with 0 signs of any problem )

asciilifeform: dulap

asciilifeform: i'ma test on dulap also, it has identical fg/cabling setup to primary s.mg box

asciilifeform: mod6: dulap is a colo, recall, it wasn't sold

asciilifeform: then , BingoBoingo : plox to power down rk 'A' and insert its stick into dulap , for reimaging, when subscr expires ☟︎

asciilifeform: mircea_popescu: asciilifeform's old kernel consists of hand-selected modules, corresponding to dulap-III irons, config can be seen here, http://nosuchlabs.com/pub/conf_current.conf ☟︎

asciilifeform: the 1 non-trinquian component is the kernel, which is bitwise identical to what is on 1) dulap 2) the primary s.mg box. fortunately kernel carries own libc, musltronicity dun affect it. but the trinquian kernel src and my original config is still there, as well as my config for the iron; it is possible for diana_coman to later build the kernel with trinque's gcc, what is needed is to transform my config so that all necessary modules

asciilifeform: currently i am not able to make use of trinque's builder ( even though i fed it my kernel config verbatim ) as it does not handle the building and installation of the ramdisk where modules live ( dulap kernel is not presently cleansed of modulism )

asciilifeform: soooooo diana_coman , mircea_popescu , i was able to bring up the box strictly by copying over a copy of my own dulap kernel

asciilifeform: BingoBoingo plox to sit down on somewhere other than dulap ip. ☟︎

asciilifeform: seems to affect at the very least dulap and the new box

asciilifeform: mircea_popescu: i've seen 0 flicker on pizarro pipes, period, since i plugged dulap in.

asciilifeform: BingoBoingo: nic on dulap was dead for a good 3 minutes.

asciilifeform: BingoBoingo: ummm didja unplug dulap ?"!?!!!

asciilifeform: ( as seen on dulap and rockchip )

asciilifeform: trinque: i'ma revv it up with my current dulap kernel, customer can reconfig as desired

asciilifeform: diana_coman: i was referencing the 'let's stand up clone of dulap' variant, strictly.

asciilifeform: diana_coman: as i understand there are exactly two variants : 1) waitloop on trinque to declare his item as ready for battlefield test 2) asciilifeform bakes an exact clone of existing s.mg gentoo ( really, dulap-iii's gentoo )

asciilifeform: mircea_popescu: i considered making a http://www.loper-os.org/?p=2295 -style canned gentoo out of my dulap trad-gentoo, but really would rather switch to trinque's item when it is baked

asciilifeform: it will be getting the asciilifeform-baked trad gentoo, from same tar.gz as dulap and orig s.mg.

asciilifeform: would take approx 3-6 mo. on dulap to break a summer-2017-and-prior infineon-generated key .

asciilifeform: mircea_popescu: ping BingoBoingo ( to plug the ssd into dulap ) and asciilifeform ( to work the magics ) if you want it relinuxed.

asciilifeform: on ch10 ffa: exponentiator test from end of ch.7 runs in 9.2s on dulap, and 28.4s on the test rk3328-roc-cc machine, producing correct output.

asciilifeform: ave1: btw the parallelism worx great, thing builds on dulap in <1hr

asciilifeform: ( for < 2x of that figure, you get a dulap-class box, with 32 cpu, 256G ram, raid card, etc )

asciilifeform: ( on dulap, it's only 900MB, incidentally )

asciilifeform: ( a build on dulap took ~5 hrs )

asciilifeform: ( iirc gentoo on dulap -- and then as supplied to s.mg -- was baked from a stage4 of same vintage )

asciilifeform: roughly speaking, the data set can double in mass one more time, before asciilifeform has to ask for a mechanical disk to be fitted to dulap ( writing 256+GB/day will annihilate ssd in a coupla months ) and get used to much slower run

asciilifeform: coincidentally, jump of www traffic on dulap in past half hr.

asciilifeform: if the volunteered proggy makes sense to me, i will emplace it on dulap and say 'thank you'

asciilifeform: from the fate of dulap-I and -II i can dare to hypothesize that they use moar cheapo local orcs who dream of miami, than 'strike team' with proverbial stainless steel staples in passport . but beyond this, nfi

asciilifeform: ^ dulap-III fg emplacement

asciilifeform: dulap II rather

asciilifeform: dulap-III humming along very happily.

asciilifeform: correct : 1, i call 'dulap-III' ; 2 spares .

asciilifeform: 1 (dulap) disked; other 2 want disks ( can be filled within 2-3 days after mp writes in re what he wants in'em )

asciilifeform: mod6: there is not some private segment that was not shown ( aside from sept.-dec. exchanges where asciilifeform was directed to obtain $iron, reported on progress, all the way to 'dulap-III ready for installation' )

asciilifeform: so what, phf is under orders not to ship completed dulap-III ? mircea_popescu proclaims 's.nsa dun need a www' or FG storefront or wat

asciilifeform: all 3 boxes purchased under s.nsa flag; #2 is provisioned with same contents as Dulap-II; the other 2 are sitting diskless and await instruction from mircea_popescu re what he'd like in'em.

asciilifeform: #2 is Dulap-III, has FG installed

asciilifeform: http://btcbase.org/log/2018-02-02#1779772 << i have 3 boxes prepared for bisp under s.nsa banners ; 1 is dulap-III , which is 'endisked', en-FG'd, and crated for its journey; 2 want disks; if mircea_popescu has specific config he wants for #2 and #3, or to obtain moar , gpggram. ☝︎

asciilifeform: even dulap-III, heavy opteron box sitting on my desk and awaiting shipment to BingoBoingoistan, is a gentoo box.

asciilifeform: dulap-II had a serious striped-raid ; still took 5x as long to verify same block as fist-sized, raidless ssd zoolag.

asciilifeform: ( dulap had ~ok trb performance on spinning rust, but it had striped raid . and STILL couldn't keep up with ssd zoolag, despite the latter being a box the size of my fist, with no raid, on residential fiber )

asciilifeform: a circa 2008 box, for >8x what i paid for a dulap-class opteron cum ssd raid

asciilifeform: http://btcbase.org/log/2018-01-03#1762790 << the last copy before destruction of dulap-II ☝︎

asciilifeform: dulap-III has only 4 fans, plus 5th in ps, none of'em any bigger than a 'D' battery. and yet sounds quite like shopvac at full blast.