[SOLVED] Cannot use Yum. Error with rpmdb?

Josh Hathcock · 03-27-2014, 04:48 PM

I am running Centos 6.5. I run

Code:

sudo yum update

I get

Code:

rpmdb: Thread/process 16272/140241829988096 failed: Thread died in Berkeley DB library
error: db3 error(-30974) from dbenv->failchk: DB_RUNRECOVERY: Fatal error, run database recovery
error: cannot open Packages index using db3 -  (-30974)
error: cannot open Packages database in /var/lib/rpm
CRITICAL:yum.main:
Error: rpmdb open failed

I have tried

Code:

sudo yum clean
sudo rm -f /var/lib/rpm/__db*
sudo rpm --rebuilddb

Removing the __db* files works.

However I then get this when I run

Code:

sudo yum update

Code:

Resolving Dependencies
There are unfinished transactions remaining. You might consider running yum-comp
lete-transaction first to finish them.
The program yum-complete-transaction is found in the yum-utils package.
--> Running transaction check
---> Package glib2.x86_64 0:2.26.1-3.el6 will be updated
---> Package glib2.x86_64 0:2.26.1-7.el6_5 will be an update
---> Package grep.x86_64 0:2.6.3-4.el6 will be updated
---> Package grep.x86_64 0:2.6.3-4.el6_5.1 will be an update
---> Package kernel.x86_64 0:2.6.32-431.11.2.el6 will be installed
---> Package selinux-policy-targeted.noarch 0:3.7.19-231.el6 will be updated
---> Package selinux-policy-targeted.noarch 0:3.7.19-231.el6_5.1 will be an upda
te

And I just get popped back out into prompt. Of course the rpm directory just gets filled with new __db* files. I noticed there is a .rpm.lock file there, which I attempted to remove as well, but it still doesn't work.

Somewhere in this process I also ran across

Code:

freeing read locks for locker rpm

Any ideas?

knudfl · 03-27-2014, 06:55 PM

Welcome to LQ.

As your text says, run # yum-complete-transaction

... But you may not be able to do that, unless `yum-utils' is already installed.
So please install "yum-utils" manually :
# rpm -Uvh yum-utils-1.1.30-14.el6.noarch.rpm
http://mirror.centos.org/centos/6.5/...el6.noarch.rpm

(( Or if all other yum related packages are updated :
# rpm -Uvh yum-utils-1.1.30-17.el6.noarch.rpm
http://mirror.centos.org/centos/6.5/...6_5.noarch.rpm ))

-

Josh Hathcock · 03-27-2014, 08:11 PM

Thank you for the reply. Indeed I wasn't able to install via yum, so I was able to install via rpm. Indeed yum-complete-transaction was successful, however now when I try to update I just get this

Code:

Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
 * base: mirror.solarvps.com
 * epel: epel.mirror.constant.com
 * extras: mirror.solarvps.com
 * remi: remi.mirrors.hostinginnederland.nl
 * rpmforge: repoforge.mirror.constant.com
 * updates: centos.aol.com
Setting up Update Process
Resolving Dependencies
--> Running transaction check
---> Package glib2.x86_64 0:2.26.1-3.el6 will be updated
---> Package glib2.x86_64 0:2.26.1-7.el6_5 will be an update
---> Package grep.x86_64 0:2.6.3-4.el6 will be updated
---> Package grep.x86_64 0:2.6.3-4.el6_5.1 will be an update
---> Package kernel.x86_64 0:2.6.32-431.11.2.el6 will be installed
---> Package selinux-policy-targeted.noarch 0:3.7.19-231.el6 will be updated
---> Package selinux-policy-targeted.noarch 0:3.7.19-231.el6_5.1 will be an update
---> Package yum-utils.noarch 0:1.1.30-14.el6 will be updated
---> Package yum-utils.noarch 0:1.1.30-17.el6_5 will be an update

It doesn't get any further along than that. I repeated my previous steps (deleting __db* files, rebuilding db, etc) and attempting to update again, and it just spits out the same results :/ No error or anything. Note that I ran yum-complete-transaction one more time and received this

Code:

Freeing read locks for locker 0x5: 16725/139639347144448
Freeing read locks for locker 0x7: 16725/139639347144448
Freeing read locks for locker 0x8: 16725/139639347144448
Freeing read locks for locker 0x9: 16725/139639347144448
Freeing read locks for locker 0xa: 16725/139639347144448
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
 * base: mirror.solarvps.com
 * epel: epel.mirror.constant.com
 * extras: mirror.solarvps.com
 * remi: remi.mirrors.hostinginnederland.nl
 * rpmforge: repoforge.mirror.constant.com
 * updates: centos.aol.com
Checking for new repos for mirrors
No unfinished transactions left.

Also .rpm.lock is back. Any thoughts?

*Edit*

Code:

yum search nodejs

returned

Code:

rpmdb: Thread/process 16748/140520319166208 failed: Thread died in Berkeley DB library
error: db3 error(-30974) from dbenv->failchk: DB_RUNRECOVERY: Fatal error, run database recovery
error: cannot open Packages index using db3 -  (-30974)
error: cannot open Packages database in /var/lib/rpm
CRITICAL:yum.main:

Error: rpmdb open failed

rknichols · 03-28-2014, 10:55 AM

Those __db* files that you keep deleting are just the transaction lock storage. It is all of the other files in that directory that are the Berkeley DB files. The "Packages" file is the critical one. All the rest are built from that.

Try this:

Code:

cd /var/lib
tar -cvf rpmdb.tar rpm                # Save the current contents, just in case...
find rpm/* ! -name Packages -delete   # Remove everything except the Packages file
rpm --rebuilddb                       # Rebuild

That should leave you with a good database. In any event, you can always restore the tar file to get back to where you were.

Josh Hathcock · 03-28-2014, 07:13 PM

Thanks for the reply rknichols. Unfortunately that didn't work. I was able to successfully able to rebuild the db, but yum still just goes acts like it's going to update, and then just pops me out to prompt.

John VV · 03-28-2014, 10:41 PM

as for fixing ?
unsure and possible not easily fixed ( if fixable )

but i think i know what caused this

Quote:

* base: mirror.solarvps.com
* epel: epel.mirror.constant.com
* extras: mirror.solarvps.com
* remi: remi.mirrors.hostinginnederland.nl
* rpmforge: repoforge.mirror.constant.com
* updates: centos.aol.com

RPMforge in incompatible with epel
and also incompatible with some things in remi

and remi is incompatible with the centos base and update -- CAUTION !!! is required

and a few packages in rpmfogre are incompatible with the cent base and update repos

remi and rpmfogre are normally DISABLED and only turned on for installing specific things

and the yum plugin "yum-priorities" is NOT installed and configured
that plugin is designed to help STOP the system from installing incompatible rpms

see the centOS wiki on installing software repos
http://wiki.centos.org/AdditionalResources/Repositories
and on that page is a link to the yum plugin
http://wiki.centos.org/PackageManagement/Yum/Priorities

installing rpms that conflict with other system critical programs can cause all kinds of problems

reinstalling might be required , while fixing rpmforge and remi can be done , it is not easy

Josh Hathcock · 03-29-2014, 10:03 AM

Thanks John VV, that makes sense. So I removed those repos and restarted my machine and voila! It works. Now, this is the first time I restarted my machine since the problem. I know, I know, that's the classic troubleshooting step, but I had ran across this problem before with another VPS which I did restart going through rebuilding the rpm db, and it didn't work. So I would be inclined to say that those repos were indeed the culprit. Thanks everyone!