Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [HTCondor-users] peaceful node drain and shutdown
- Date: Mon, 08 Aug 2016 15:04:43 -0400
- From: Michael V Pelletier <Michael.V.Pelletier@xxxxxxxxxxxx>
- Subject: Re: [HTCondor-users] peaceful node drain and shutdown
To pick up a discussion that left off in mid-July...
A cryptic error crops up in condor_q when the START expression is set to
UNDEFINED or ERROR when running an analyze:
condor1$ condor_q -better-analyze 264
-- Schedd: condor1 : <138.127.79.182:1567?...
error: bad form
error: problem with ExprToProfile
User priority for pelletm@cade is not available, attempting to analyze
without it.
---
This appears to arise in boolExpr.cpp from ExprToMultiProfile() - it's
complaining that the Start value as a Boolean type is not either True or
False.
So it would appear that the proper approach to refusing additional jobs
while keeping existing jobs running, while also avoiding having condor_q
complain to the user about something outside of their purview, is to sever
START and IS_OWNER as suggested, via START = False along with IS_OWNER =
False, rather than setting START to UNDEFINED or ERROR.
But in thinking about this further, it seems like this is roughly
equivalent to draining a machine with an infinite retirement time. Is
there any way to initiate draining through ClassAd expressions, I wonder,
along the lines of "START_BACKFILL = True?" That is, START_DRAINING =
True?
-Michael Pelletier.
_