pcp
[Top] [All Lists]

Re: [pcp] debian package installation failure

To: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Subject: Re: [pcp] debian package installation failure
From: Nathan Scott <nathans@xxxxxxxxxx>
Date: Thu, 13 Feb 2014 17:16:50 -0500 (EST)
Cc: PCP Mailing List <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <52FD3EDB.5030503@xxxxxxxxxxxxxxxx>
References: <52FD3EDB.5030503@xxxxxxxxxxxxxxxx>
Reply-to: Nathan Scott <nathans@xxxxxxxxxx>
Thread-index: v0E0CED51SS1NhmlwqDEMQ2hJcsu/Q==
Thread-topic: debian package installation failure

----- Original Message -----
> After the latest packaging changes ...
> 
> This used to work just fine:
> 
> kenj@bozo:~/src/pcp$ sudo dpkg -i build/deb/*.deb
> 
> But now ...
> [ok stuff deleted]
> Unpacking replacement pcp ...
> dpkg: warning: unable to delete old directory '/etc/pcp/pmwebd': Directory
> not empty
> dpkg: warning: unable to delete old directory '/var/log/pcp/pmwebd':
> Directory not empty
> 
> Which is not good.

Its the best that can be done AFAICT, no configs or logs are lost & the
new package then assumes responsibility for these paths, as it should.

> Setting up pcp (3.9.0) ...
> Installing new version of config file /etc/init.d/pmproxy ...
> Installing new version of config file /etc/init.d/pmie ...
> update-rc.d: /etc/init.d/pmmgr: file does not exist

This is the root cause.  Its a leftover reference in the pcp postinst
script, and its exit code causes abortion of the entire install.  Just
a bit odd that this didn't fail for me - anyway, its fixed in git now.

> And then I get a whole slew of what look like consequential errors because
> pcp 3.9.0 did not get configured ...

Yep, everything else flows from that.  For me it installs cleanly - it
may be I'm upgrading from 3.9.0 -> 3.9.0 though, and so missed this in
the process of rearranging the furniture.

cheers.

--
Nathan

<Prev in Thread] Current Thread [Next in Thread>