1.3.4 - high cpu load - client login

[Florent]

@Tom-Elliott Yes imagereplicatorglobalenabled is not checked

[Florent]

@UWPVIOLATOR said in 1.3.4 - high cpu load - client login:

Increased MaxRequestWorkers from 150 to 500 in mpm_prefork_module
Hi,
Where you have put this parameter because i try in my apache2.conf in prefork module but not valid after i restart apache2

[ablohowiak]

So no improvement with imagereplicatorglobalenabled off?

With this issue I’ve increase my client communication up to 30 minutes. I will reduce when I know things have stabilized.

In Ubuntu, see which mpm you have enabled. There should only be one in /etc/apache2/mods-enabled. That’s the file you would edit to increase the MaxRequestWorkers. Make a copy of the file before editing.

[Florent]

@ablohowiak said in 1.3.4 - high cpu load - client login:

So no improvement with imagereplicatorglobalenabled off?

No

With this issue I’ve increase my client communication up to 30 minutes. >I will reduce when I know things have stabilized.

Where you have up to 30 minutes your client communication , “checkin time parameter” ?

[ablohowiak]

Yes.

[Florent]

@ablohowiak
Oh very thanks it’s works !
I was thinking test this before but maybe not with “imagereplicatorfogenabled” at off.
After few minutes cpu is ok.

Thanks.

[Florent]

@ablohowiak @Tom-Elliott

Hi,

Today it’s better but all the 30 minutes we have pic of 2 minutes.

In wiki it’s write (https://wiki.fogproject.org/wiki/index.php?title=FOG_Client

The frequency of the checkin-time determines how quickly the FOG Client will receive instructions from the FOG Server. If an image deployment is scheduled for a computer that is turned on, with a checkin-time of 60 seconds, means the FOG Client may begin initiating the task anywhere from 0 to 60 seconds + the random staggering time that is added. This same concept would apply to immediate power management tasks, snapin tasks, capture tasks, and so on. Scheduled tasks are not affected by this behavior, and if the target system is on when the scheduled task is to be ran, this will happen on time

How it’s possible to modify the random staggering range for spread over 10 minutes for example ?

Thanks.

[Wayne Workman]

@Florent said in 1.3.4 - high cpu load - client login:

How it’s possible to modify the random staggering range for spread over 10 minutes for example ?

The staggering is not meant to be adjustable, what’s meant to be adjustable is the FOG_CLIENT_CHECKIN_TIME, this governs how often the FOG Client should check with the server. Assuming 10,000 computers power on all at exactly the same moment, over the course of the day the random staggering will cause all clients to spread out evenly with their check-ins. But nobody has 10,000 systems all powering on at the same second. Generally, organizations either A. leave their computers on all the time or B. allow the end-users to power them on.

Also asking @Joe-Schmitt if he could give further explanation.

[ablohowiak]

1.3.5-RC 10 and working fine.

All clients are back on, I’ve already reduced the client check in time, and the load was fine for the start of the school day.

The Image replicator service is still off.

[Florent]

@Wayne-Workman
Hi,
Ok i understand.
Actually majority off our classroom reboot at the same time for software update in night (same time is residual effect with green fog on old version).
I will adjust power management (good amelioration in this version) for spread out this reboot.

Thanks.