asciilifeform: re turks, supposedly 80+% of ottoman-era dictionary consisted of persisms/arabisms.
asciilifeform: compared to what, e.g., mao, did to ~his~ lang -- very small diff.
asciilifeform: sorta funny , in 1990s ru various 'monarchist'-style folx went around trying to put'em back ~in~ . but naturally did not know the proper rules for their use, ended up with somewhat comical output
asciilifeform: but currently i suspect it aint the stick, watch #p for updates
asciilifeform: mircea_popescu: sticks dun have smart ( at least any i've seen )
asciilifeform: however, diana_coman's unit boots, and initializes ssh , etc , but doesn't appear to actually listen on the port
asciilifeform: the sad boxen boot if stick is moved to usb2 jack. i currently suspect loose or contaminated contacts in usb3 ( it has additional pins towards the back )
asciilifeform: we've discovered some pretty odd things
asciilifeform: diana_coman: plz give ( or not ) permission to modify your rk filesystem to enable uart console ( /etc/inittab )
asciilifeform: diana_coman: can you ssh to your rk nao ?
asciilifeform: ( which is usb2 only ) so i currently suspect dirty or underforced connections
asciilifeform: one of the 'dead' sticks seems to read entirely ok in dulap
asciilifeform: diana_coman: plz pop into #p & watch for updates, we are in the process of eliminating other potential culprits
asciilifeform: diana_coman: this is strictly optional
asciilifeform: and why asciilifeform did not hurry to buy 9000 of these drives
asciilifeform: now you folx know what 'pilot plant' means !
asciilifeform: ( there are two spares on hand, for this event )
asciilifeform: diana_coman: i'ma ask BingoBoingo to plug this disk into one of your other boxes (plox to choose which one) so you can recover any state you had not a chance to back up; meanwhile you will be issued a spare .
asciilifeform: [ 9.105595] blk_update_request: I/O error, dev sda, sector 33827664 << consistently across boots
asciilifeform: i'ma add it to the template, it will go on all new/reimaged boxen.
asciilifeform: the rk's are on own physical switch, so if one of them pulls the lan to ground, it will take out only rk cluster. but that's the only isolation available.
asciilifeform: as the iron currently stands, it is a hostel, not hotel, and folx must refrain from shitting in one another's beds.
asciilifeform: written after 1st time we had this thread.
asciilifeform: because there is, in effect, 1 shared wire.
asciilifeform: mircea_popescu: right nao anybody can sit down on as many ips as he wants, or munge and retransmit , or pretty much any conceivable mischief.
asciilifeform: mircea_popescu: if you know how to perma-fix this with a box that doesn't cost 50k or take up an entire rack (i.e. x86 box with 9000 nics) i'm all ears.
asciilifeform: l1 folx who do this, will be publicly shamed. new customer -- booted.
asciilifeform: mircea_popescu: it is only 'resolved' in the sense you are contemplating, on juniper and similar.
asciilifeform: they're distinguished by mac ( which incidentally a malicious user could easily reset to be whatever he wants ) so the problem is only solvable in the general case by a router which operates jacks individually
asciilifeform: this requires 'smart' switch, where individual jacks are addressible
asciilifeform: i do not currently know of any practical means to entirely isolate the boxes on the lan.
asciilifeform: mircea_popescu: i warned of this in the first days ( currently digging in log re exact line )
asciilifeform: and unfortunately it is the case that ips are static, and set in box, and accessible by user, this was in the log when the plant went online.
asciilifeform: mircea_popescu: i test'em before handing over to mod6 .☟︎
asciilifeform: ACHTUNG all pizarro rockchip customers ! if you modified your /etc/conf.d/net , in recent 2 weeks, plz say ! asciilifeform currently suspects that somebody munged theirs, and it may account for diana_coman's unreachable box, we may have to take the pilot plant down for maintenance and manually dig, if nobody says anyffing
