@user17915 There is a webmaster SE where it might be more on topic.

@FaheemMitha I'd rather not copy and paste this on other sites. Is the chat there active?

@user17915 I agree with @FaheemMitha, another site may be better suited than SO (which is a programming site). You should delete your question from SO and re-post it on the most relevant site. The chat for that site should be available in a similar ways as this one is for U&L.

@user17915 That's not an error, so if it's failing to start it's for a different reason

@user17915 You should check first whether it's on topic on the other site.

I don't know about the chat. I'll check. One sec.

The only room linked to Webmasters is chat.stackexchange.com/rooms/524/webmasters

Last post 13 days ago.

@user17915 Your error should probably be formatted separate, rather than being inline.

And it looks more like a warning than an error. I suppose you did the customary searches?

@FaheemMitha may be it is, I am still learning this so still don't know much about it to say either way, but it's preventing me from restarting the httpd service. But the problem seems to be the ErrorLog /var/www/html/username1/error.log line in the VirtualHosts config file. The system is preventing me from creating file inside /var/www/html/username1/.

@FaheemMitha It's something related to file permissions and I'm not sure why just following the instructions on that tutorial doesn't work. As a temporary measure I can comment the ErrorLog line to get this to work though

@user17915 So the server won't restart? Then it's an error.

Not sure how to put this as a good question, but if I accidentally start hundreds of compute intensive processes, is the computer supposed to hang so bad the only thing left to do is hard reset it?

It was a for loop in bash where I shouldn't have put a & after the command in the loop body

@Nobody I don't know about "supposed", but it might.

Though that is more likely to reflect I/O than CPU issues.

But it sounds like you might be using the wrong tool. Though without more details alternative suggestions are shots in the dark.

@FaheemMitha I started an imagemagick convert job on a bunch of files, so yes, it's possible there would be IO problems, although the time for the process should be like 5s read, 100s compute, 5s write. The next time I just let it run sequentially or used sem -j +0 to manage parallelism, but I was surprised I could crash my computer by just doing it naively - I expected to at least be able to kill the terminal emulator.

There is also imagemagick mogrify for batch jobs, but somehow I didn't get that to work.

@Nobody You probably ran out of RAM. If you let it run, the OOM killer would eventually reap the processes and the machine would come back, but yeah, you can certainly freeze your system by starting too many processes. That's what a fork bomb is, after all.

@terdon Possible. I did wait half an hour or so. The input and output file would fit into RAM more than 10 times, but possibly imagemagick does something stupid or there are other inefficiences (the algorithm I ran should be possible to run in-place, I think, it was a -morphology Close).

@Nobody Provided you have swap, you would notice that you were running out of memory, because your machine would start swapping.

Under those circumstances, it would not freeze completely, just become much less responsive.

If you don't have swap, then yes, it could freeze completely so that it would require a hard reboot.

@FaheemMitha I have 8GB of RAM and 8GB of swap. When I ran it I saw about 20 background processes being launched with the ids listed in the terminal, then everything froze.

It was pretty much instant.

@Nobody Hmm. That does sound odd.

So the keyboard and mouse were completely unresponsive?

Yes. I unsuccessfully tried switching to a tty and the mouse pointer wouldn't move either

It's a laptop keyboard and I don't think that special kernel sysrq combo exists here so I couldn't try that.

@FaheemMitha my experience is that it's terribly hard to recover from thrashing with a runaway process. And gnome rarely survives...but that's largely gnome's fault

@Nobody if you mean REISUB, it does work sometimes

@Nobody When under high load, the most probable console that would work is one of the six at CTRL-ALT-F1 to F6. Those are consoles with very low resources consumption. You can log in there and kill some/many/all of the runaway processes. Those consoles are available almost anywhere.

@Nobody If the computer is so loaded that not even such consoles are available, the next option (if it was configured to be available) is CTRL-ALT-BACKSPACE to kill X and all processes that X has as child's.

@Nobody The issue with REISUB is that you should have tried before hand as the combination of keys needed to activate it may be odd/strange.

@Nobody You can find out which parts are active by cat /proc/sys/kernel/sysrq (on Linux) and by reading what is explained here: kernel.org/doc/html/latest/admin-guide/sysrq.html

@Nobody On the laptop I am using now, the combination is FN-ALT-SHIFT-SYSRQ-H to get help. Change the last letter to get all other sysrq commands.