Re: [HTCondor-users] why does htcondor change sysctl params, and why is this done outside of /etc/sysctl.{d, conf} ?

[SCHAER Frederic <frederic.schaer@xxxxxx>]

Hi *, and thanks for your replies,

Mmm... I did not mean to start a controversy.
Actually, I didn't know about sysctl.d not being part of LSB neither, but anyway, I consider it being good practice.

I do use puppet like others.
And like others, I've noticed /etc/sysctl.d was supported since at least RHEL5, and even that the apply_sysctl() function is applied by and only by the network init script in RHEL<=6 (don't know about what's applying it in RHEL7, but directory is there, indeed)

Anyway : I can perfectly understand the rationale for managing hundreds of thousands of jobs (... millions ? ), and indeed I can adapt my monitoring to this by lowering the frequency of tests.
By the way : the monitoring I am speaking of looks for phalanx, and if you're a phalanx victim, you cannot rely on the presence of the malware processes in /proc.

So, to be back on this : may I suggest that if sysctl.d directory exists, a .txt file is put in there by the condor default tuning script if that's the one that's used ?
I'm assuming that if/when people change the default tuning script, they will know what they are doing, and nothing has to be done then by condor.

There is one thing I disagree though : if the script/condor is "entitled" to tune the running kernel, I do believe it is also entitled to create a potentially missing system directory (or a lambda directory in /etc from the LSB point of view ;) ) - after all, if you can touch the heart of the sytem, you can probably touch the brain, what's the difference ?
If directory does not exist... the tuning script can probably create it too and put a .txt inside, as this will cause no harm and the script must have root privileges to tune the kernel anyway.

Another suggestion : since this kernel tuning only seems valid for the linux platform as the documentation suggests, would it be possible that the default script logs the kernel tuning it does in the syslogs  (via the logger command) in addition to any other logging mechanism ?
I have the feeling that such system changes belong to be logged in the syslogs too, not just condor logs ?

Frederic

-----Message d'origine-----
De : HTCondor-users [mailto:htcondor-users-bounces@xxxxxxxxxxx] De la part de Rich Pieri
Envoyé : vendredi 15 janvier 2016 01:29
À : htcondor-users@xxxxxxxxxxx
Objet : Re: [HTCondor-users] why does htcondor change sysctl params, and why is this done outside of /etc/sysctl.{d, conf} ?

On 1/14/2016 5:13 PM, Dimitri Maziuk wrote:
> Since I have not created anywhere, it's all stock centos, I have to ask:
> it does not exist on which exactly "standard RHEL and its derivatives"?

Support for sysctl.d was added to RHEL 5 and 6 in Spring 2011 but
neither create the directory out of the box. RHEL 7 is the first version
to do so.

-- 
Rich Pieri <ratinox@xxxxxxx>
MIT Laboratory for Nuclear Science
_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/htcondor-users/