Weblog entry #108 for Steve
See all weblog entries made by Steve.
There are some mistakes which I frequently make, despite knowing better. These are almost always the result of my fingers being faster than my brain; and that isn't agood trait.
The worst thing is the moment when you realise you've made the same mistake before...
With the intention of curing myself of these stupid lapses heres my list of mistakes I frequently make:
- Typing "killall ..." on Solaris machines.
- Rather than killing all processes with the name "..." this kills all processes. Reboot time.
- Getting the arguments to nice / renice the wrong way round - e.g. "renice -19 -u skx"
- I'd only use nice if the load was very high; so getting the argument the wrong way round and allowing the errant proceses to use more CPU time makes things worse, not better.
I'll update this entry with more as I remember them.
So what do you do wrong most often?
Comments on this Entry
1) It was time to go to bed and I wanted to shut down my workstation. For me, the procedure to shutdown was :
$ su Password: # shutdown -h now
But I had forgoten that I was logged on the server. So those commands shut the server down.
So now I have two things for this :
- I turn off the workstation using a point and click menu
- I changed the root prompt of the server to include a big warning.
2) I tarred something (using tar cvzf dir/*) and I forget the archive name, which destroys the last file in the bash expansion.
[ Parent | Reply to this comment ]
# killall^W uname -a
[ Parent | Reply to this comment ]
s/pavlonian/pavlovian/
[ Parent | Reply to this comment ]
Generally mistakes are deleting things by accident but hey thats what backups are for right
[ Parent | Reply to this comment ]
Oh yeah, and `chattr +i` might be your friend too ;-)
Laters.
[ Parent | Reply to this comment ]
[ Send Message | View Steve's Scratchpad | View Weblogs ]
It would definitely be nice to have that kind of thing, although thankfully recovering /bin isn't so hard.
I managed to recover my system without too much panic!
[ Parent | Reply to this comment ]
Probably not the place to ask I guess, but I'm here already ;-)
[ Parent | Reply to this comment ]
[ Send Message | View Steve's Scratchpad | View Weblogs ]
I guess the really simple approach would be to get the source to the busybox-static package, and rebuild that to install into a custom directory.
That should give you plenty of tools - I've been working on something kinda similar to this for work, there I also build the mdadm tools, and the LVM tools - so we can boot RAID + LVM systems with completely static tools.
Although it has to be said that building the package would be the easy part. Getting the package added to the project should be easy enough if there was a good rationale for adding a new top-level directory. (Filesystem Hierarchy Standards people might have comments? But actually adding it as a base/required (compulsory) package would be controversal, especially when the busybox-static package is already available for people who want this kind of software available...
[ Parent | Reply to this comment ]
That should be OK, except this time apache was upgraded (and thus stopped), causing anyterm (it's a terminal via the webserver, ssh is blocked here) to fail, and thus causing the whole upgrade to stop.
This left the system without running apache, so unfixable without ssh access:(.
(Still, anyterm is really useful for those behind firewalls that block ssh on port 22 and the https port)
[ Parent | Reply to this comment ]