I think I understand why Linux doesn't implement SIGINFO (Ctrl+T) - the system stops responding because you were extracting a 800MB xz compressed file on a machine with 8 cores and 16GB ram. (It's been 25 minutes now...)
What good is siginfo when you can neither switch windows, move the cursor nor press any keys?
Very pragmatic
Do I wait longer? (I don't think the command should take this long?) Or do I reboot and risk corrupting the filesystem? What is the OS doing spinning the fan at full speed all this while? I can't even run top, I hit the shortcut to open new console tab 10 minutes ago...
Switching away from windowing to system console took a while, and over a minute to switch back. Not trying that again.
Much Windows 98 nostalgia happening X-)
It definitely shouldn't take 40 minutes to extract a file on Linux. Hard reboot it is. Wish me luck with the filesystem not crapping its pants.
Heh, I thought I'd be clever and run a FreeBSD VM to automate testing some code before pushing to my home server. We're still stuck extracting the VM image file for thrice longer than it would take me to SSH into my server and run every thing by hand.
In unrelated tech woes, wanting to shave off tens of kilobytes from memory requirement on ESP32 devices while running Micropython (freezing the custom modules into the firmware) involves either downloading docker images (nope, not in this economy) or... gigabytes of source code(?) by cloning the ESP-IDF repos and building the firmware locally.
I mean, just the WiFi repo is:
Cloning into '~/esp/esp-idf/components/esp_wifi/lib'...
remote: Enumerating objects: 54762, done.
remote: Total 54762 (delta 0), reused 0 (delta 0), pack-reused 54762 (from 2)
Receiving objects: 100% (54762/54762), 1.06 GiB | 1.86 MiB/s, done.
LOLWHUT?
A total of 4.5 GB to build a ~1MB firmware. Software building toolchains are
(All this to afford me that 1MB interpreter/firmware to build tools with shared code for myself whether on FreeBSD, MacOS,Linux or ESP32/8266 devices. Before one DeBloats, one must embody TheBloat.)
Realised just now** that I've been poking at this (build custom micropython on unix, esp8266 and esp32) for nearly 12 hours.
**because the Makefile finally works for all targets!
@hiway IME this happens when you have too much swap. Something starts runaway allocation and swapping is excruciatingly slow because it happens on page granularity, synchronously. My view is that you should never have more swap than you'd be willing to wait for copying with dd, 4k block size, sync after each block.
@dalias I haven't tweaked anything when installing this system, it is as the distribution creators have set it up :')
@hiway Yeah, nobody seems to be able to agree on good defaults...
But there's some runaway resource usage bug triggering your problem anyway. Swap behavior just allows it to get nasty bogged down rather than failing early.
@dalias I'll read up on how to shrink the swap. The RAM usage rarely goes above 6GB in my daily use (have 16), so swap/thrashing never crossed my mind. Thank you :)
@dalias the system had 2GB swap, temporarily turned it off completely. I'm curious if that was also causing the intermittent lockups (and not just this one long freeze) when the laptop had uptime greater than a handful days.
@anandphilipc yup, yup!
@anandphilipc hahah! It is just as true X-)
@hiway if you don't need the history, consider cloning with depth 1? ...or is it 1gb even with a shallow clone?
@greg good point! I should try that in a clean env soon - the instructions are to init git submodules recursively (and run some scripts that the toolchains provide which do this automatically for deeper nested submodules) - not sure if I can safely override those without getting too much in the weeds.
@hiway
If the machine is that frozen then you have other issues. The kernel should never be so hard hung that some of that simple stuff can't happen.
Half the use of siginfo for me is to see if the kernel is frozen or not (as the kernel line should appear near instantly), and let me know what wait state the current process is in (eg. is it swapping?).
@encthenet I've turned off swap on advice from another fedizen, let's see how that fares. Although my RAM use is barely 6GB out of 16 on any usual day.