TESP-API configuration #230
Conversation
group_vars/galaxyservers.yml
Outdated
| assign: ['db-skip-locked'] | ||
| execution: | ||
| default: tpv_dispatcher | ||
| default: tesp_env |
There was a problem hiding this comment.
this changes the executor for everything, we want to only send specific users to tesp (e.g. myself), we need to change tpv configuration to achieve this
There was a problem hiding this comment.
Would you be able to configure that for the usegalaxy.cz testing ?
There was a problem hiding this comment.
Yes, the instances have independent tpv config files, look at https://github.com/CESNET/usegalaxy/blob/main/files/usegalaxy.cz/tpv_rules_local.yml
| require: | ||
| - tes | ||
| destinations: | ||
| - id: tesp_env |
There was a problem hiding this comment.
IMHO, this couldn't work. AFAIK, you cannot use outside-TPV destinations (environments now) inside TPV. You need to specify an inside-TPV environment which will require a specific tag (e.g., tesp) and then use this tag to limit people or tools, etc.
|
Not quite sure about TES specific parameters, but the structure should be fine now |
| 32346539383333363333333661663232313663626433346232623837643231356234653834323261 | ||
| 30326639666163313530616363316135653565663639396631373863613532356234613237656537 | ||
| 62383831663335633865623933636230353966653266643939646164646538333664 | ||
| 35346536366566333666616433613161373130313134623465633832376438343965363264393036 |
There was a problem hiding this comment.
what is being changed in the vault?
| remote_metadata: false | ||
| rewrite_parameters: true | ||
| outputs_to_working_directory: false | ||
| submit_native_specification: "-l select=1:ncpus={int(cores)}:mem={int(mem)}gb:scratch_local={int(scratch)}gb -l walltime={int(walltime)}:00:00 -q {{ pulsar.pbs_queue }} -N {{ pulsar.nfs_prefix }}_j{job.id}__{tool.id if '/' not in tool.id else tool.id.split('/')[-2]+'_v'+tool.id.split('/')[-1]}__{user.username if user and hasattr(user, 'username') else 'anonymous'}" |
There was a problem hiding this comment.
In this line of code, it is very important to understand that user and pbs queue used to submit jobs is set in the playbook individually for each Galaxy server and it is connected to the dual Pulsar playbook configuration. In other words, if TES Pulsar does run under a different user than user configured here it likely will not work. These two users must be the same. Therefore we need to know which user is submitting jobs to PBS.
3 participants
Configuration to run jobs on TESP-API server.