asciilifeform: each will sit on individual rail clamp.
asciilifeform: BingoBoingo: re 'fingerwork' -- i've dismounted the rk's for the 1u earlier built for transport to piz , and they will live in a 'DIN' 2u subrack. so no moar finger gymnastics.
asciilifeform: asciilifeform-rk hosting is a 'turn-key' product, incl. gentooism, maintenance, debuggery as-req'd, etc
asciilifeform has no intention of attempting gentoo port to 'raspberry' or any other 'boot blobs'-laden arm machine
asciilifeform: BingoBoingo: funnily enuff the upstream provider colos (not leases) 'raspberries' for 10.
asciilifeform: BingoBoingo: i'd like to keep prices 'as low as possible, but not lower'. heathendom offers much lower, theoretically, prices, than what is seen in this prospectus . and not erryone agrees ' asciilifeform's hands worth added 50/mo ', so keeping the greed to a min.
asciilifeform: a 256G 'dulap' potentially could house ~128 users at similar ram apportionment (but only 8 if each were to be allotted 4 cores as present in rk.)
asciilifeform: BingoBoingo: this is so. the current standard rk contains 2GB ram.
asciilifeform: BingoBoingo: i think you know, i had rk provisioning ~100% automated.
asciilifeform spent 4d nao thinking about 'how to price', 'what to price', and head hurts.
asciilifeform: so we'll defo want the provisions for it.
asciilifeform: BingoBoingo: for l2 n00bs can do shared hosting , per '1y phree' scheme suggested earlier by mp.
asciilifeform: incl. 'bouncers', i want that 'ddos proof' claim tested sooner rather than later.
asciilifeform: BingoBoingo: as for rk vs shared hosting, i actually prefer that folx live in rk, it is considerably easier to administer , each user gets actual physical machine where can do as he pleases.
asciilifeform: BingoBoingo: i did not want to start right off the bat w/ explicit pipe rationing, because it in itself imposes some loss of total capacity. but the provisions for rationing are in place.
asciilifeform: BingoBoingo: shares-of-pipe are to be handled as described under 'physical space & energy' # III .
asciilifeform: BingoBoingo: i omitted discussion of shared hosting strictly because the provisions do not yet exist. but it should be simple to compute the cost of emplacing the necessary machine, via this draft.
asciilifeform: all prospective inhabitants , invited to comment.
asciilifeform: ^ a mechanized price estimator is included .
asciilifeform: then cannot make excuse 'oh, spamfilter ate.'
asciilifeform: BingoBoingo: since you have a printer, may be worth sending same to their snail addr. notarized.
asciilifeform: BingoBoingo: ianal -- but seems like admission of invalidity of contract. if they intend to persist in the claim to the oct-feb moneys, would then logically continue to sit on the irons, neh
asciilifeform: BingoBoingo: lol, maybe 1% of it ( what was left trapped in that chest ? coupla rk ? )
asciilifeform meanwhile cooked up the spreadshits ; tonight after salt mine will write article re costs.
asciilifeform: diana_coman: presently occupied with ^ .
asciilifeform: diana_coman: that thing is on insulator pedestal, under it is a 3mm steel sliding plinth. the temp in the dc is a comfortable ~20C when i last measured. (surprisingly, very quiet vs latech's, they have civilized freon cooling, entire back lot fulla chiller towers. and, interestingly, coupla thou. litres of genuine pre-ban halon 1301 for fire suppression.)
asciilifeform: diana_coman: the move will involve ~30 seconds of down time, we will schedule it together prior to any movement.
asciilifeform: diana_coman: currently your box looks like this. it'll be sat down on iron DIN rail w/ additional forced air when i put this in, it is still snailing its way through the mails.
asciilifeform: i'ma carry out test where iron can be ruled out as limiting reactant, as soon as i get chance.
asciilifeform: on a good day i got ~2MB/s from dulap. (peak).
asciilifeform: diana_coman: didja ever make a note of what sorta throughput same item had at piz ?
asciilifeform: i'ma put in 'apu1' w/ primo samsung msata later this wk, for logger, will make it available for similar test.
asciilifeform: this appears to be rate-limited by rk ssd.
asciilifeform: diana_coman: before you go back to these -- is there a large file i can wget-test with from here , in ossasepia ? i'm curious re whether oceanic pipe is limit, or the speed of the rk ssd
asciilifeform: i was curious of-what it was made tho.
asciilifeform: mp_en_viaje: right, but seems like there's an ancient never-flushed cache somewhere ?
asciilifeform: unrelated q for iptronix gurus : wai does traceroute to diana_coman's www , show a long-dead rubbish domain on last hop ? ( netsideproductions.com ) ? is this config issue somewhere ? (afaik harmless, but i'd like to know why it happens)
