[Hcoop-discuss] regarding admin's and responsibilities

[Adam Chlipala]

Sorry for the delayed response.  I've been traveling with limited 
Internet access.

I agree with Davor that adding more admins is not a solution, though 
Tanveer probably didn't mean to suggest that.  We need to come up with 
some more precise criteria for the duties of an admin, and then find 
people who can meet these criteria:

- Schedules of guaranteed time commited to HCoop.  We should find an 
admin line-up that has reasonably uniform coverage of time periods, and 
each admin should have a posted schedule that he is obligated to adhere 
to.  Even if we have spotty coverage, making the schedules official is a 
big improvement.
- Guarantee of minimum 24-hour turn-around on any HCoop-related e-mail.  
All admins should be reasonably easy to contact by e-mail.  (I think I 
maintained this property for myself even while I was traveling, handling 
any unhandled support requests I found during my daily e-mail check.)
- Official notice of any period when these duties can't be fulfilled.  
If anything comes up that makes an admin too busy to fulfill any one of 
his duties, then he should provide official notice of what time period 
this will cover, so that we can re-shuffle schedules to ensure adequate 
coverage.  I think it is reasonable to waive this requirement only for 
the most dire and unexpected situations.
- Ability to handle any issue.  It makes sense to assign a primary admin 
to deal with non-time-critical issues about each service, but everyone 
with root access should be able to handle any emergency.

Some other issues:

Given what I believe is the wide appeal of HCoop, I don't think there's 
any reason to settle for any admin who doesn't have substantial 
professional sysadmin experience.

When we determine exactly what admin coverage we need, we can figure out 
what, if any, additional compensation to admins is necessary to maintain 
enough competence and dedication.

I am not a sysadmin.  I don't know many basic things about system 
administration, and I don't want to learn... but I think I've still had 
and can continue to have much to contribute so far as software 
infrastructure goes.  This means that it's useful to have input from 
experienced sysadmins (and others) related to how to implement new 
HCoop-specific infrastructure.  I'd like to have a more open process for 
deciding on technical policies.  Ideally, every decision should be 
documented on the wiki, and the whole mass of decisions should be 
relatively easy to navigate through.  Every new decision should be made 
after a discussion on this mailing list, and from time to time we should 
have outsiders perform a complete audit of our policies and procedures.  
It's probably time to do this last thing relatively soon, though 
possibly after moving to a new primary server set-up.