From stonjek@fnal.gov Tue Oct 19 16:36:10 2004 Date: Thu, 14 Oct 2004 16:58:54 -0500 (CDT) From: Stefan Stonjek To: Richard St Denis Cc: Art Kreymer Subject: List of CDF SAM machines and what they do Hello, here is a list of all CDF SAM nodes I can think of and what they are doing (are supposed to do :-)) I think that this list is complete (Yes I know that Art uses cdfpcb for installation test, but I think it should not be included into this list). Tschuess Stefan ************* Stefan Stonjek *************** stonjek@fnal.gov ********** ***** Fermilab ; MS 318 ; P.O. Box 500 ; Batavia, IL 60510, USA ***** ******** phone: +1-630-840-2969 ********** fax: +1-630-840-8859 ******** * Univ. of Oxford; Physics Dept.; Oxford; OX1 3RH; UK; +44-1865-273442 * List of CDF SAM machines and what they do ========================================= cdfsam01 -------- - naming service for productions - including cron job: name_service_cleanup - db-servers for for production - 4x version 5 - 1x version 6 (multi threaded) cdfsam02 -------- - naming service for integration - db-servers for for integration - 3x version 5 - 1x version 6 (multi threaded) - naming service for development - db-servers for for development - 3x version 5 - 1x version 6 (multi threaded) fcdfdata016 ----------- - SAM station for CAF - name of SAM station: cdf-sam - this station uses dCache fcdfdata014 ----------- - SAM station for remote storing of files to enstrore - name of SAM station: cdf-cat cdfsamint --------- - SAM station in the integration universe - name cdf-int cdfsamdev --------- - SAM station in the development universe - name cdf-dev cdfsamth -------- - SAM station for the test harness fcdfsun1 -------- - web-server - this include cron jobs which copy the web pages once an hour from the CVS respository and cron jobs which generate the consumption plots. - log-server (all this should move to the new web server as soon a we get the hardware) ncdf151 ------- - machine fore SamTV - located in Randy's office (ncdf109, my personal desktop which run a cronjob: /cdf/home/stonjek/bin/define-datasets.sh) --------------------------------------------------------------------- We already ordered new hardware. The machines will be all the same, except for the web server which will have a bigger disk. cdfsam03: production dbserver with new code for gradual switchover and unanticipated fast demand from offsite growth. We dont have any failover as we have actually stolen cdfsam02 to implement the new dbserver. We realized that it is powerful to be able to put up a dbserver we think is ready to use, run jobs against it and then switch over sites gradually. We can even do this on a "CAF process type" designation (ie. sam-test). fcdfdata01x : production sam station for gradual switchover Partner to match the switchover philosophy mentioned above. We have stolen cdfsamint (see below) for this purpose. We want to be able to move this to be the production target -0- and then use the other machine (016) for the next phase of deployment. cdfsam04: web pages server, log server, big disk (140G mirrored) This is currently being done by fcdfsun1 and we want to put that horse out of its misery. cdfsam05: SamTV/monitoring (separate out from web pages to make sure a crash of the monitoring does not bring down web services) Currently being done by Randy's ncdf151 machine. You decide if we steal it permanently or not.