[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] docker detection broken in centos 7



The starter does not massage the output, it captures it into a set of 8Kb buffers using block reads
and then prints those buffers out one line at a time. 

piping the output of "docker info" to a file would be a good approximation of what the starter does.

If the total output exceeds 8Kb, then I might want to look for bugs in the way we handle lines that
straddle the buffer boundary, but other than that we don't filter the output.  what shows up in
the starter log when D_FULLDEBUG is configured is what we get from the command.

-tj


-----Original Message-----
From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> On Behalf Of Dimitri Maziuk via HTCondor-users
Sent: Friday, November 1, 2019 1:59 PM
To: htcondor-users@xxxxxxxxxxx
Cc: Dimitri Maziuk <dmaziuk@xxxxxxxxxxxxx>
Subject: Re: [HTCondor-users] docker detection broken in centos 7

On 11/1/19 1:37 PM, Greg Thain wrote:

> If you run /usr/bin/docker info as root, do you get the warning? Also,
> what does ls -ld /root/.docker say?

/root/.docker doesn't exist, never has,

CLI output is the same for root and condor user,

there are other differnces between CLI output and StarterLog.

If you're not massaging it in starter, my best guess is something in
shell environment makes `docker info` behave differently than when
exec()'ed?

-- 
Dimitri Maziuk
Programmer/sysadmin
BioMagResBank, UW-Madison -- http://www.bmrb.wisc.edu