DS Project Issues

Staffing

iRODS

replication

processes

Only one dedicated person

No power to get people to do things

DS admin members getting conflicting priorities between DS and team they belong to, err on the side of the person that signs their paychecks

management

too many unfinished/postponed tasks

I'm bad at estimation

impedance mismatch between CyVerse WBS and Development tasks

lots of bugs. I have 33 bug workarounds in our iRODS configuration and rule logic

poor tooling for administration. I've written over 20 scripts to help with administration tasks

dynamic PEPs don't align well with actions of interest, very poor documentation

log files are poor in 4.2

way too easy for users to a cause a mess

no locking

data-store-fix shouldn't exist, don't know why it is fixing 10s of issues a day.

TACC environment is different

Unavailable for months at time when upgrading hardware

I should have to fix 100s of issues every time I replicate files to TACC

current process is too manual

testing

VCS branching strategy too complicated

hard to test iRODS rules without deploying them, slow

no gatekeeping for what's deployed. I've been surprised a few times by something showing up in prod

dev local test env is brittle

ansible+python+centos == headache

little monitoring

very little documentation