System hardware recommendations

I cant find where you guys recommend how much ram or diskspace. We have to order some some new boxes and I was looking for what you would recommend. We are going to running Linux boxes with Gui and this is for the master schedulers running about 100 -150 jobs a day. I know the requirements are probally not that high but i tried running it on one of our old test boxes that was using windows 2k with 512k ram and the JAVA kept hanging up on us.

Hope the following info can help you to estimate space and memory requirements

The scheduler files take about 12 MB, JRE is about 40-50 MB depending on platform, if all the "extra" stuff is stripped it comes down to less than 30 MB.
The job database file is relatively small - it is a compressed file typically in 20 - 200 KB size range. The largest file in terms of file size I've seen so far was about 450 KB and contained over 300 jobs with large JavaScripts scripts, job comments, etc...

To run jobs, the scheduler also allocates space in queue subfolders which are used only while jobs are being run. It requires about 1 to 2 KB per job instance. So if you run 50 concurrent jobs (or 50 instances of the same job), you need 50 to 100 KB for job queues. However, if you hit a case of a runaway job and nobody is watching the system, a lot of pending jobs can accumulate in the same queue until a preset queue limit is hit. For that bad scenario, you may need to have another 1 or 2 MB of space available.

The total size of log files depends on the scheduler settings and typically varies from 100 KB to several MB. Yet, I've seen some log files as big as 20 - 30 MB

All together, this comes to about 55 to 80 MB of space. Of course, this number doesn't include any external files and programs installed on the system which are run from jobs.

For memory requirements, it is recommended to have 1 GB on the scheduler system and 512 MB or more on agent systems, if agents are used. The total required RAM is highly dependent on what jobs are doing, if they are memory hungry and how many of such jobs could run concurrently. Theoretically it is possible to create a JavaScript job in 24x7 loading a lot of data into memory arrays and that may require a lot of memory, but that is of course a radical case.