Groupe des Utilisateurs Altiris Suisses et Francophones Group

Console Win32 d’Altiris Deployment Solution 6.9 sp3

Certains opérateurs de la console win32 DS « glissent » un groupe de 125 jobs sur un groupe de 78 ordinateurs pour réinstaller des salles de cours:


Le problème est que cette manipulation provoque un Timeout de la console Altiris Deployment :

 

<o p=""> </o>

Nous sommes obligés de redémarrer les services Altiris sur le serveur de déploiement pour débloquer la situation et les jobs s’effectuent de manière anarchique sur les postes clients.
- Ce qui présente un réel problème, car en général, on souhaite des installations dans un ordre bien déterminé...

En observant le serveur de déploiement, on ne voit pas de charge particulière durant cette manipulation.<o p="">
</o>

<o p="">
Problème relevé à L'université de Fribourg: </o>hubert.clement

Avez-vous déjà eu vent d’un tel problème ?
Existe-t-il un nombre max. de jobs que l’on peut installer en même temps ?

Cher Pascal,

L'article ci-dessous fournit une réponse à cette épineuse question. Encore merci pour ton intervention et amicales salutations
Hubert

Applies To

• Deployment Solution 5.6, 6.0, 6.1, 6.5, 6.8, 6.9

Question

How many resources can be targeted with a scheduled job at any one time before the Deployment Server starts to have problems?

Answer

The actual bottleneck for scheduling jobs is the communication between the Deployment Solution console and the SQL server. Usually if you schedule a job like this against a large number of computers all at once, it will create a SQL deadlock. Another bottleneck is when the axengine is actually sending out the jobs to the client computers.   We really can’t give a specific number since there are so many variables. If you had a Deployment Server system where you had four quad core processors (16 total logical processors) with 16 GB of RAM, the SQL server was local, and there were 15 hard disks in 3 RAID volumes (for the SQL server, eXpress share, and operating system), you would probably be able to schedule a really large number of jobs at once. We’ve rarely seen a Deployment Server system like that. On an average system you wouldn’t be able to schedule quite as much jobs. Also, some eXpress databases are kept fairly clean while some have bulked up with a lot of extra garbage data. This can affect the way Deployment Server handles jobs.   Some Deployment Solution server systems could probably handle scheduling 1500 jobs at once while other Deployment Server systems could not. A definitive number is difficult to give because of the number of variables and really it can’t be given at a theoretical level. All of the current suggested load limits that are published in our documents are gathered from consultants who set up our systems at customer sites and then report back what they experienced when implementing our product.   The bottom line is that there have been customers that could only schedule about 20–30 jobs at once with their Deployment Solution system (any more and their SQL server would lock up) and we’ve seen customers that have successfully scheduled 4000 jobs at once. There is no published limit because it really varies so much on what is set up in their environment and how it is set up.   The suggested best practice is to schedule the jobs to run in batches of no more that 200 at a time. However, this figure also depends on the type of job being run. If the job has multiple tasks then this figure would need to be decreased.