Discussion Forum: Thread 324784 |
|
|
| | Author: | Fr0stByt3 | Posted: | Jul 29, 2022 15:10 | Subject: | systems engineering problems in a nutshell | Viewed: | 241 times | Topic: | Off Topic | |
|
| Imagine there being a slew of tools set to a specific group environment that
you need to newgrp into (yes, this is a Linux server). Now imagine you break
a registry somewhere by running a script that aids in domain replication. Can't
figure out what the script exactly did but all you know is that now you can't
newgrp anymore, and so every specialized tool you use is inaccessible.
So now you have this registry you're trying to kill by its process ID. Sending
a SIGKILL signal to it after doing ps -ef just closes your terminal. Logging
back in, it's still there. You really need to kill, cycle, and restart that
registry before proceeding with diagnosis and repair. Best close the domain and
then the service in charge of making sure things start up automatically.
Can't. Both ksh scripts you have available to stop the domain require the
registry working in order to proceed. Might as well restore from a snapshot at
this point.
Messy.
|
|
|
| | | | | |
| | | | Author: | 1001bricks | Posted: | Jul 29, 2022 15:12 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 53 times | Topic: | Off Topic | |
|
| SomeRandomThoughts?
|
|
| | | | | | | | | |
| | | | | | Author: | Fr0stByt3 | Posted: | Jul 29, 2022 15:18 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 65 times | Topic: | Off Topic | |
|
| In Off Topic, 1001bricks writes:
| SomeRandomThoughts?
|
Pretty much.
Too bad that this is an actual problem I'm facing currently. I should document
the issue and my procedures next time to make sure I can backtrack if needed
and have a guide for people to follow if I fix it and they end up having the
same issue down the road.
|
|
| | | | | | | | | | | | | |
| | | | | | | | Author: | miskox | Posted: | Jul 29, 2022 16:36 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 47 times | Topic: | Off Topic | |
|
| Replace Linux with OpenVMS? Problems solved.
Saso
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | Author: | 1001bricks | Posted: | Jul 29, 2022 17:16 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 44 times | Topic: | Off Topic | |
|
| In Off Topic, miskox writes:
| Replace Linux with OpenVMS? Problems solved.
|
Read in a TV manual translation: "In case of any problem, please first unplug
the technical support".
I swear!
|
|
| | | | | | | | | | | | | | | | | | | | | |
| | | | | | | | | | | | Author: | BricksOfFaith | Posted: | Jul 29, 2022 17:24 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 38 times | Topic: | Off Topic | |
|
| In Off Topic, 1001bricks writes:
| In Off Topic, miskox writes:
| Replace Linux with OpenVMS? Problems solved.
|
Read in a TV manual translation: "In case of any problem, please first unplug
the technical support".
I swear!
|
Tech support still says everything is fine!!! I’m telling you!
|
|
| | | | | | | | | | | | | | | | | |
| | | | | | | | | | Author: | peregrinator | Posted: | Jul 29, 2022 22:48 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 42 times | Topic: | Off Topic | |
|
| In Off Topic, miskox writes:
| Replace Linux with OpenVMS? Problems solved.
|
Man I miss VMS. Those were the days.
|
|
| | | | | |
| | | | Author: | BricksOfFaith | Posted: | Jul 29, 2022 17:00 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 41 times | Topic: | Off Topic | |
|
| In Off Topic, SomeRandomUser writes:
| Imagine there being a slew of tools set to a specific group environment that
you need to newgrp into (yes, this is a Linux server). Now imagine you break
a registry somewhere by running a script that aids in domain replication. Can't
figure out what the script exactly did but all you know is that now you can't
newgrp anymore, and so every specialized tool you use is inaccessible.
So now you have this registry you're trying to kill by its process ID. Sending
a SIGKILL signal to it after doing ps -ef just closes your terminal. Logging
back in, it's still there. You really need to kill, cycle, and restart that
registry before proceeding with diagnosis and repair. Best close the domain and
then the service in charge of making sure things start up automatically.
Can't. Both ksh scripts you have available to stop the domain require the
registry working in order to proceed. Might as well restore from a snapshot at
this point.
Messy.
|
Tech support says everything is fine! 😂
|
|
|
| | | | | | | | | |
| | | | | | Author: | Fr0stByt3 | Posted: | Jul 29, 2022 17:10 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 34 times | Topic: | Off Topic | |
|
| In Off Topic, BricksOfFaith writes:
| Tech support says everything is fine! 😂
|
Yep. All is fine. Nothing responding on the front end? Perfectly normal.
|
|
| | | | | |
| | | | Author: | qwertyboy | Posted: | Jul 29, 2022 17:47 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 38 times | Topic: | Off Topic | |
|
| In Off Topic, SomeRandomUser writes:
| Imagine there being a slew of tools set to a specific group environment that
you need to newgrp into (yes, this is a Linux server). Now imagine you break
a registry somewhere by running a script that aids in domain replication. Can't
figure out what the script exactly did but all you know is that now you can't
newgrp anymore, and so every specialized tool you use is inaccessible.
So now you have this registry you're trying to kill by its process ID. Sending
a SIGKILL signal to it after doing ps -ef just closes your terminal. Logging
back in, it's still there. You really need to kill, cycle, and restart that
registry before proceeding with diagnosis and repair. Best close the domain and
then the service in charge of making sure things start up automatically.
Can't. Both ksh scripts you have available to stop the domain require the
registry working in order to proceed. Might as well restore from a snapshot at
this point.
Messy.
|
This all sounds weird. There is no registry in Linux (a registry is a Windows
thing). A Linux server can run Windows domain services through Samba, but this
is all configured in txt files - mainly smb.conf, or sometimes in a database
(but that would be specified in your smb.conf).
A SIGKILL won't terminate your shell, unless you either kill the shell itself,
kill its dependency (like an ssh session), or try to do something like `kill
-9 -1`.
If you need to use `newgrp` this implies you need to change the user's default
group membership for that current session. Is that user actually allowed to change
group membership (see `man newgrp`)? Does `newgrp` actually spit out a message
(or, failing that, what is its exit value)? If `newgrp` fails, you can also simply
(as root) use `vipw`/`vigr` to change that user's default group, and then
log in as that user.
Niek.
|
|
|
| | | | | | | | | |
| | | | | | Author: | Fr0stByt3 | Posted: | Jul 29, 2022 17:58 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 48 times | Topic: | Off Topic | |
|
| In Off Topic, qwertyboy writes:
| This all sounds weird. There is no registry in Linux (a registry is a Windows
thing). A Linux server can run Windows domain services through Samba, but this
is all configured in txt files - mainly smb.conf, or sometimes in a database
(but that would be specified in your smb.conf).
A SIGKILL won't terminate your shell, unless you either kill the shell itself,
kill its dependency (like an ssh session), or try to do something like `kill
-9 -1`.
If you need to use `newgrp` this implies you need to change the user's default
group membership for that current session. Is that user actually allowed to change
group membership (see `man newgrp`)? Does `newgrp` actually spit out a message
(or, failing that, what is its exit value)? If `newgrp` fails, you can also simply
(as root) use `vipw`/`vigr` to change that user's default group, and then
log in as that user.
Niek.
|
To be fair, I'm a fledgling systems engineer undergoing training. I was using
a simulated learning environment with nodes designed to mimic what would be found
in the corporate production domains. The terminology is still relatively new
to me.
Performing a kill -9 on the registry was actually killing the monitor. Corporate
tools are assigned to a special Linux group and breaking the corporate registry
by running a script led to the newgrp command (done by typing in "newgrp - ")
saying there were too many arguments among other things.
I'm still learning. Only been at this a month and I'm grateful they taught
me like it was a classroom setting. The domain replication is what gave me severe
trouble.
|
|
|
| | | | | | | | | | | | | |
| | | | | | | | Author: | qwertyboy | Posted: | Jul 29, 2022 18:19 | Subject: | Re: systems engineering problems in a nutshell | Viewed: | 49 times | Topic: | Off Topic | |
|
| In Off Topic, SomeRandomUser writes:
| To be fair, I'm a fledgling systems engineer undergoing training. I was using
a simulated learning environment with nodes designed to mimic what would be found
in the corporate production domains. The terminology is still relatively new
to me.
|
Fair enough. But using terminology that doesn't match 'real life'
will make it hard for others to help you.
| Performing a kill -9 on the registry was actually killing the monitor. Corporate
tools are assigned to a special Linux group and breaking the corporate registry
by running a script led to the newgrp command (done by typing in "newgrp - [group]")
saying there were too many arguments among other things.
|
`newgrp - [group]` is not a valid command. You either want to change your primary
group (and use `newgrp [group]`) or reset your shell to its defaults (and use
`newgrp -`). Specifying both `-` and `[group]` is not allowed, and `newgrp` tells
you that - there are too many arguments.
| I'm still learning. Only been at this a month and I'm grateful they taught
me like it was a classroom setting. The domain replication is what gave me severe
trouble.
|
Breaking a test environment and then trying to right the ship is one of the ways
you can learn the most. Try to approach your issue using small steps, and don't
try to fix it "in one go". Use tools that allow you to poke at your environment
in a controlled manner (like `smbclient`, `testparm` etc). Compare the setup
with a known working one (especially smb.conf). GUIs and scripted solutions can
hide the actual problems, especially ones which are not foreseen by its authors.
Good luck!
Niek.
|
|
|
|
|
|
|