Bug #2792: During a migration from 2.3.8 to 2.4.0~beta3, rudder-agent postinst can display some errors when copying CFEngine binaries - Rudder - Issue Tracker

Actions

Copy link

Bug #2792

closed

During a migration from 2.3.8 to 2.4.0~beta3, rudder-agent postinst can display some errors when copying CFEngine binaries

Added by Nicolas PERRON over 11 years ago. Updated about 9 years ago.

Status:

Released

Priority:

Assignee:

Jonathan CLARKE

Category:

Packaging

Target version:

2.4.0~beta5

Pull Request:

Severity:

UX impact:

User visibility:

Effort required:

Priority:

Name check:

Fix check:

Regression:

Description

[...]
rudder-agent              0:off  1:off  2:on   3:on   4:on   5:on   6:off
cp: cannot create regular file `/var/rudder/cfengine-community/bin/cf-execd': Text file busy
cp: cannot create regular file `/var/rudder/cfengine-community/bin/cf-serverd': Text file busy
rudder-agent has been installed, but not started
[...]

It could be because of Rudder Webapp which uses cf-promises before generating any promise.

We should use a better test before to copy any CFEngine binary.

Related issues 1 (0 open — 1 closed)

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

Status changed from New to Pending technical review
% Done changed from 0 to 100

Applied in changeset commit:6aaa38e3ee60a2755872c69b0179f6da8b0805bb.

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

To ensure that no CFEngine process is up, we kill all /var/rudder/cfengine-community/bin binaries.

Actions

Copy link

Updated by Matthieu CERDA over 11 years ago

OK, with NPE correction it looks much more correct :) Approved !

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

Status changed from Pending technical review to In progress
% Done changed from 100 to 90

It seems that the correction doesn't work. The CFEngine binaries aren't copied.

I'm on it

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

I can't understand. I've tried to debug this with additionnal outputs and it works:

rudder-agent              0:off  1:off  2:on   3:on   4:on   5:on   6:off
list CFEngine process
use pkill in order to free CFEngine binaries
Exit status: 1
Try to copy binaries
Exit status: 0
rudder-agent has been installed, but not started
Making a key pair for cfengine, please wait, this could take a minute...

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

The error during installation and migration was coming from an exit error of pgrep, pkill and grep -v when these commands didn't find any results.

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

Status changed from In progress to Pending technical review
% Done changed from 90 to 100

Actions

Copy link

Updated by Nicolas PERRON over 11 years ago

Status changed from Pending technical review to In progress
Target version changed from 2.4.0~beta3 to 2.4.0~beta4
% Done changed from 100 to 90

This is not a blocking issue so I change it to the next run. We still need technical review for the work done.

Actions

Copy link

Updated by Matthieu CERDA over 11 years ago

Looks OK. But about the errors that cp outputs, a simple "cp ~~a /opt/rudder/sbin/cf~~* /var/rudder/cfengine-community/bin/ >/dev/null 2>&1" might make it shut up.

Actions

Copy link

#10

Updated by Nicolas PERRON over 11 years ago

The problem has not been fixed so we still need work on it. Technical review can't be considered as done.

Actions

Copy link

#11

Updated by Jonathan CLARKE over 11 years ago

Assignee changed from Nicolas PERRON to Jonathan CLARKE
Target version changed from 2.4.0~beta4 to 2.4.0~beta5

I'll take a look at this. Maybe do a retry if it fails once?

Actions

Copy link

#12

Updated by Jonathan CLARKE over 11 years ago

Target version changed from 2.4.0~beta5 to 2.4.0~rc1

Actions

Copy link

#13

Updated by Jonathan CLARKE over 11 years ago

I think the problem here is not how you stop the CFEngine processes, just that you can't stop them this way. As a reminder, there is a cron job that restarts CFEngine daemons every 5 minutes if they're stopped... so you can kill them anyway you like (pkill, killall, kill -KILL), they still may be started up again by another script.

I think a good workaround here would be to disable any programs from using the binaries, via a mechanism such as described in #2884.

Actions

Copy link

#14