SEE-GRID OPS role
From EGEE-see WIki
This page describes the configuration of the OPS role for SEE-GRID sites. OPS role is intended to be used for monitoring so SFT jobs could be distinguished and have priority over normal (application) jobs.
[edit]
OPS role in SEE-GRID VOMS
Mihajlo has been added to this role as he is responsible for sending SAM jobs to SEE-GRID sites:
https://voms.irb.hr:8443/voms/seegrid/webui/admin/users/list?rolename=Role%3Dops
Proxy for this role is created using the following command:
voms-proxy-init -voms seegrid:/Role=ops
[edit]
Site configuration
- Add the new user to users.conf (changing the IDs and names as appropriate for your site):
2018:seegridops:1400:seegrid:seegrid:prd:
Instead of seegridops you can also use seegridprd user (but probably not both).
- Add the new role to group.conf (prd tag is used to link the entries in this two files ). It is not possible to use ops as a tag because older versions of yaim only support sgm and prd tags.
"/VO=seegrid/GROUP=/seegrid/ROLE=sgmadmin":::sgm: "/VO=seegrid/GROUP=/seegrid/ROLE=ops":::prd: "/VO=seegrid/GROUP=/seegrid/*"::::
- Make sure you have the following in site-info.def:
VO_SEEGRID_VOMS_SERVERS="vomss://voms.irb.hr:8443/voms/seegrid?/seegrid/"
- Reconfigure CE, SE, RB, LFC, VOBOX or at least run:
/opt/glite/yaim/scripts/run_function site-info.def config_users config_mkgridmap
- Reconfigure WN or at least run:
/opt/glite/yaim/scripts/run_function site-info.def config_users
- Check the user:
$ id seegridops uid=2018(seegridops) gid=1400(seegrid) groups=1400(seegrid)
- Check the mappings:
$ grep seegrid /opt/edg/etc/edg-mkgridmap.conf # Map VO members (Role) seegridsgm group vomss://voms.irb.hr:8443/voms/seegrid?/seegrid/Role=sgmadmin seegridsgm # Map VO members (Role) seegridops group vomss://voms.irb.hr:8443/voms/seegrid?/seegrid/Role=ops seegridops # Map VO members (root Group) seegrid group vomss://voms.irb.hr:8443/voms/seegrid?/seegrid .seegrid
$ grep seegrid /opt/edg/etc/lcmaps/gridmapfile "/VO=seegrid/GROUP=/seegrid/ROLE=sgmadmin/Capability=NULL" seegridsgm "/VO=seegrid/GROUP=/seegrid/ROLE=sgmadmin" seegridsgm "/VO=seegrid/GROUP=/seegrid/ROLE=ops/Capability=NULL" seegridops "/VO=seegrid/GROUP=/seegrid/ROLE=ops" seegridops "/VO=seegrid/GROUP=/seegrid/Role=NULL/Capability=NULL" .seegrid "/VO=seegrid/GROUP=/seegrid" .seegrid
- Assign additional priority for seegridops user, perhaps by allowing it to use the dedicated CPU. Following maui.cfg snippet assignes one CPU to be used only for dteam, ops and seegridops jobs:
# use any working node SRCFG[monitoring] FLAGS=SPACEFLEX # reserve one cpu SRCFG[monitoring] TASKCOUNT=1 RESOURCES=PROCS:1 # reserve it forever SRCFG[monitoring] PERIOD=INFINITY # it can be used by seegrid ops user SRCFG[monitoring] GROUPLIST=seegridops # give it enough priority to run imediately GROUPWEIGHT 1 GROUPCFG[seegridops] PRIORITY=1000000 # or by egee dteam and ops queues SRCFG[monitoring] CLASSLIST=dteam,ops # give them enough priority to run imediately CLASSWEIGHT 1 CLASSCFG[ops] PRIORITY=1000000 CLASSCFG[dteam] PRIORITY=1000000
