RPM Community Forums

Mailing List Message of <rpm-users>

Re: About the consistency of /var/lib/rpm/__db.00?

From: Jeff Johnson <n3npq@mac.com>
Date: Tue 25 May 2010 - 23:08:16 CEST
Message-id: <C1A03860-45BC-4FD8-93D4-84A6172B9FA9@mac.com>

On May 25, 2010, at 5:01 PM, Marc MERLIN wrote:

> 
>> Note that running (*DB->verify) is directly at odds with
>> "not happening" I/O as a goal as well.
> 
> Yep. I didn't actually need verify, so I removed it. 
> "private" is really the only useful bit.
> 

The consequence of "private" is that its _ENTIRELY_
up to you, not RPM, to ensure rpmdb access serialization. I've
been publically accused of "recommending" eliminating rpmdb
locking as a solution in the past, so

	YOU HAVE BEEN WARNED!!!!!!!!

	Do you understand the consequences of your decision (yN)?

But there's also a global fcntl lock "lockfd" on Packages
that can be resurrected if necessary too. What's screwy
is that RPM sets the lock, but doesn't honor the lock, so
a 1-2 line logic change (in rpmdb/db3.c, look for fcntl) i
 needed if you wish RPM to do global exclusive locking.

hth

73 de Jeff
Received on Tue May 25 23:08:47 2010
Driven by Jeff Johnson and the RPM project team.
Hosted by OpenPKG and Ralf S. Engelschall.
Powered by FreeBSD and OpenPKG.