/srv/irclogs.ubuntu.com/2021/08/29/#ubuntu-discuss.txt

lotuspsychjegood morning02:39
Oxfuxxx_Hi.04:25
lotuspsychjemorning Oxfuxxx_ 04:44
=== EriC^^_ is now known as EriC^^
TJ-away10:56
TJ-Does apt have a parallel download add on? something said yesterday made me think we have all this parallel processing now but the package downloads are mostly serialised, when apt could be fetching from multiple mirror servers at the same time11:01
lotuspsychjesounds like something the mirror guys could answer TJ- 11:05
TJ-I doubt it; probably juliank 11:11
TJ-hmmm, debootstrap failing early without errors, after successfully downloading packages but has Packages file named as debootstrap.invalid_dists_bionic_Release and so on :s11:24
TJ-ahh, that isn't a bug, that's part of how it names the manually fetched Package/Release files; so what's going on!? Time to dig the code11:28
TJ-ahhh "Failure trying to run: chroot /target dpkg ...11:28
lotuspsychjedont crack your brain on sundays TJ- lol11:32
TJ-seems like I'm not being given the choice!11:32
TJ-wow! It is extracting the base packages (including dpkg) but the files aren't in the file-system!!)11:33
TJ-hmm! "no entry data.tar.gz in archive" doing "ar -p /path/to/base-files....deb data.tar.gz"11:44
TJ-ohhh! they're .xz not .gz inside the .deb file11:46
TJ-"ar -t /path/to/file.deb" reports content11:46
TJ-solved those by manually using ar, unxz, tar to extract base-files, dpkg, core-utils, libc6 ...only to hit "FATAL: kernel too old"12:07
daftykinsjust what are you up to? :D12:09
daftykinsor what are you cooking up today? 12:09
TJ-exploring ways to do a side-by-side 18.04 install on a 10.04 install. User in #linux brought the issue to us last week. Has an isolated 10.04 PAE with 10GiB RAM required to run old proprietary enterprise software. Can't do a VM, tool old to do proper containers, so I'm exploring if we can get a chroot 18.04 alongside that they need to run more modern tools12:12
TJ-So I've got a Pentium3 with 10GiB RAM VM guest with 10.04 i386 on it and figuring it out12:13
daftykinssheesh sounds like a mess of a scenario that only employment can present12:13
TJ-it's quite fascinating 12:15
TJ-I've told debootstrap to fetch the 18.04 kernel packages; now I'm manually extracting those on the host and installing them there; A reboot should then solve the issue12:15
TJ-that did it; 10.04 with 4.15 kernel now12:25
daftykinshave you learnt what their funky enterprise software does?12:26
TJ-internal custom ERP tied to some specialised equipment12:27
daftykinsand money is the barrier to an upgrade to modernise the platform perhaps?12:27
TJ-if systems work there's no need to replace them12:28
TJ-this modern fascination with constantly replacing something that works fine is an abberation12:28
daftykinsaside from those cases when something is full of holes perhaps12:29
TJ-When you've got equipment that had a CapEx of > £500k you're not going to scrap it just so an attached PC can upgrade software to stuff you don't need12:30
daftykinsyeah that makes sense, obviously don't know the full story here - doesn't make it any less of a pig to handle12:30
TJ-it's not a pig; good to exercise the brain cells and realise how much we break things12:31
TJ-this kind of knowledge is invaluable for data recovery, forensics, and so on12:31
TJ-thank goodness I can do it on a VM :)12:32
daftykinswell it's more difficult and not straightforward, that's what i meant by pig12:32
TJ-It makes me want to get my 2003 era Sony Vaio notebooks that have a max of 384MiB RAM running; they ran Windows XP brilliantly; I recall sitting in Starbucks branches in the USA and video-conferencing back to my UK offices when they had problems without an issue.12:33
TJ-SRX-51P SRX-41P I think are the model nymbers. Cost aroudn £2k each at the time, were the bees knees12:34
daftykinsme and my wealthy client used to really like their ultraportable range back then12:35
TJ-Yes; I still have then and all the extras in crate; been loathe to throw them away 12:36
TJ-would be interesting to see what I can get them to do..will need to replace the battery cells I bet.12:36
TJ-thankfully removable battery packs with regular sized cells inside, not LiPo packs12:37
TJ-I'll try installing Edgy or Feisty on them :)12:37
daftykinsa client says i can have this beast the next time i drop by! https://i.imgur.com/cxhYYVf.jpg12:40
TJ-is that one of the PCG- range?12:41
daftykinsyep12:41
TJ-wow, can get new batteries for the SRX51p still for under £3012:42
daftykinsok found the original shots, PCG-N505SN12:42
TJ-very surprised12:42
TJ-I've still got a Toshiba PDA from late 90s that I used to have for SatNav and other stuff, with a CompactFlash GPS card 12:43
TJ-and it still snappier than modern 'smart' phones 12:43
daftykins:D i've just boxed up my Psion Series 3c to send to a YouTuber, Neil from RMC12:44
TJ-I regret scrapping several Silicon Graphics workstations 12:45
TJ-but they were massive, heavy, and just darn great paperweights12:45
TJ-plus they had to have CRTs 12:45
daftykinsah there are tonnes of funky mods for old systems now, to take old analog output and have even low end Pi's process them and spit out over HDMI, pretty neat12:47
TJ-I did buy a convertor for those but it never worked correctly,even to VGA12:47
TJ-Those SG Indy's had very unusual video outputs12:48
daftykinsi wanted my Dad to dig out the old Apple II and Amiga to see if they still run, but he seems reluctant12:51
TJ-Here's an annoying issue. Because I'm manually extracting .debs to get it to a state where chroot will work to do the rest, the /var/lib/dpkg/status isn't being update. I /know/ I wrot a tool to do this years ago... but no idea where to find it!! Got to rewrite it12:52
TJ-I've still got my 1st Sinclair ZX81 from 1981 and it works12:52
daftykinshrmm12:53
daftykinsyeah i presume neither had an RTC back then so there may be no batteries to have leaked and ruined them12:54
TJ-rebuildoing dpkg/status is easy; "echo '' > ./empty";  "ar -x path/to/.deb control.tar.{xz,gz}"; "tar -xf control.tar.{xz,gz}"; then "cat ./control empty >> ./var/lob/dpkg/status" (./empty file makes the blank line between each package entry in 'status')13:15
=== Droid is now known as Mekaneck
=== EriC^^_ is now known as EriC^^
escanorgood name eric15:23
lotuspsychjeTJ-: https://www.reddit.com/r/Ubuntu/comments/pdz0pd/i_was_an_avid_ubuntu_user_since_gutsy_gibbon/17:05
TJ-well! nice to find out I broke my router's boot config :D19:29
daftykinsnothing is safe at TJ HQ!19:31
TJ-hehehe ! router started playing up earlier somehow broken Path MTU/MSS - still not sure if it was underlying Starlink issue or not (not tested yet) but rebooted router as a 'quick' test to find it got stuck at grub rescue> and there was no ($prefix)/boot/ 19:33
TJ-turns out, a couple weeks ago when I created the 'power-cut-proof' LUKS PXE reboot, and added a DRDB layer in betwen luks and the file system I neglected to configure it so GRUB could still read the files19:34
TJ-luckliy I'd copied /boot/ to /boot_orig/ and was able to alter the prefix and paths to vmlinuz/initrd.img to boot-strap it19:35
TJ-drat! MTU issues still19:35
Bashing-omBut TJ- Can Do It ! :D19:36
* TJ- isn't so sure!19:36
TJ-this path-MTU issue is VERY weird; only started earlier this afternoon; been fine for months19:37
Bashing-omWhen all else fails: Read the logs :P19:39
TJ-There's a wireguard tunnel over Starlink to a datacenter server where our Internet gateway is. Tunnel only carries IPv6. MTU on tunnel is 1420. On router's tunnel ingress got an ip6tables rule to clamp MSS to path MTU which means clients here doing TCP learn the correct MSS via ICMP messages when they send the initial SYN packet and get the first ACK back19:39
TJ-but suddenly that 1420 is causing failures; I've had to manually reduce it on my laptop to 1360 to get it to work. On the router where it is still 1420, TCP connections time out19:41
Bashing-omTJ-: Males one wonder what is configured at the distant end datacenter server :(19:44
TJ-I'm just wondering that myself. It's Linode London. I've just noticed the initial SYN from my local router sets MSS=1360 (weird since I've not set that on the router, only on my laptop, so it must have 'learned' it due to packets from my laptop being forwarded through) then the resulting SYN,ACK has MSS=1440 ! That's 20 larger than the interface that packet is arriving from (the public link19:47
TJ-eth0 MTU=1420)19:47
TJ-and its now fixed itself; looks like a temporary glitch inside Linode network20:07
Bashing-om\o/20:12
TJ-Apparently it can be a sign of DDoS mitigation where, when the hoster/ISP network is under attack, they trasparently tunnel the traffic through a DDoS scrubber like Cloudflare20:15
TJ-and such tunnel would reduce the MTU but presumably they force-set the TCP MSS which is why I saw 1440 on incoming packets and that ewas larger than my 1420 and those packets got dropped when trying to enter my wireguard tunnel20:16
TJ-IPv6, unlike IPv4, doesn't allow packet fragmentation so it (should) send an ICMPv6 Destination Unreachable (Packet too big)20:17

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!