Background: We produce a big Library Management System, the server parts written in C, compiled on Linux SLES 15 and deployed to ~100 customers. The version in question was compiled on SLES 15 SP2 a year ago, and our Internal IT Department updated meanwhile the Dev and QA hosts to SP3.
It turned out, that the libcrypt.so moved with this update from SP2 to SP3 to a new location, from /lib64 to /usr/lib64 and contains a new symbol:
strings /usr/lib64/libcrypt.so.1.1.0 | grep XCRYPT_2.0
XCRYPT_2.0
# rpm -q -f /usr/lib64/libcrypt.so.1
libcrypt1-4.4.15-150300.4.2.41.x86_64
# zypper info libcrypt1
Information for package libcrypt1:
----------------------------------
Repository : SLE-Module-Basesystem15-SP3-Updates
Name : libcrypt1
Version : 4.4.15-150300.4.2.41
Arch : x86_64
If you now compile a server application on SP3 and ship this to customers (as a fix for an urgent bug) who is still using SP2, these application are missing this symbol and do not start anymore:
/opt/lib/sisis/avserver/batch/bin/prg/BASTVL: /lib64/libcrypt.so.1: version `XCRYPT_2.0' not found (required by /opt/lib/sisis/avserver/batch/bin/prg/BASTVL)
# strings /lib64/libcrypt.so.1 | grep XCR
# strings /usr/lib64/libcrypt.so.1 | grep XCR
strings: '/usr/lib64/libcrypt.so.1': No such file
# rpm -q -f /lib64/libcrypt.so.1
glibc-2.26-13.48.1.x86_64
# rpm -q -f /usr/lib64/libcrypt.so.1
error: file /usr/lib64/libcrypt.so.1: No such file or directory
i.e. our internal update from SP2 to SP3, make it impossible to deliver fixes to customers running SP2, or they need update as well to SP3 before installing fixes, at least if libcrypt.so is involved.
Any comments or hints for a workaround?
At the end I compiled from source with
git clone https://github.com/besser82/libxcrypt.git
cd libxcrypt
./autogen.sh
./configure --prefix /usr/local/sisis-pap/libxcrypt
make
sudo make install
ls -l /usr/local/sisis-pap/libxcrypt/lib64
insgesamt 1300
-rw-r--r-- 1 root root 635620 26. Jul 14:09 libcrypt.a
-rwxr-xr-x 1 root root 945 26. Jul 14:09 libcrypt.la
lrwxrwxrwx 1 root root 17 26. Jul 14:09 libcrypt.so -> libcrypt.so.1.1.0
lrwxrwxrwx 1 root root 17 26. Jul 14:09 libcrypt.so.1 -> libcrypt.so.1.1.0
-rwxr-xr-x 1 root root 681656 26. Jul 14:09 libcrypt.so.1.1.0
lrwxrwxrwx 1 root root 10 26. Jul 14:09 libowcrypt.a -> libcrypt.a
lrwxrwxrwx 1 root root 11 26. Jul 14:09 libowcrypt.so -> libcrypt.so
lrwxrwxrwx 1 root root 13 26. Jul 14:09 libowcrypt.so.1 -> libcrypt.so.1
lrwxrwxrwx 1 root root 10 26. Jul 14:09 libxcrypt.a -> libcrypt.a
lrwxrwxrwx 1 root root 11 26. Jul 14:09 libxcrypt.so -> libcrypt.so
and pointed our application via LD_LIBRARY_PATH to use this version of libcrypt.so.1.
Related
On Ubuntu 19.10, I have a broken install/removal of MariaDB. I attempted every solution to this sort of problem. Yet I still can't replace MariaDB with mysql-server and am seemingly unable to remove MariaDB.
What is left on the system in /var/lib/dpkg/info/mariadb* that I want to go away is:
-rw-r--r-- 1 root root 118 Oct 27 14:02 mariadb-client-10.3.list
-rwxr-xr-x 1 root root 174 Aug 2 10:53 mariadb-client-10.3.postrm
-rw-r--r-- 1 root root 28 Oct 27 14:02 mariadb-common.list
-rwxr-xr-x 1 root root 361 Aug 2 10:53 mariadb-common.postrm
-rw-r--r-- 1 root root 509 Oct 27 14:02 mariadb-server-10.3.list
-rwxr-xr-x 1 root root 3449 Aug 2 10:53 mariadb-server-10.3.postrm
Yet when I try anything like:
sudo mv /var/lib/dpkg/info/mariadb* /tmp/
sudo dpkg --remove --force-remove-reinstreq mariadb (or variations)
gives only:
dpkg: warning: ignoring request to remove mariadb-server which isn't installed
SO how do I rebuild the database of installed packages??? and create a health system again??
I'm writing some loadable modules for Zabbix, as such, compiling shared objects. I've written one which uses the MySQL C API to read some data from tables, it's fairly standard, and includes:
#include <my_global.h>
#include <mysql.h>
My gcc command looks like so (expanded mysql_config for clarity):
gcc -fPIC -shared -o zbx_mysql.so zbx_mysql.c -I/usr/lib64/mysql `mysql_config --cflags` -I/opt/zabbix/3.2/include -L/usr/lib64/mysql -lmysqlclient -lpthread -lm -lrt -ldl
Contents of /usr/lib64/mysql:
-rw-r--r-- 1 root root 21358968 Sep 13 17:15 libmysqlclient.a
lrwxrwxrwx 1 root root 20 Nov 19 23:19 libmysqlclient_r.so.18 -> libmysqlclient.so.18
lrwxrwxrwx 1 root root 24 Nov 19 23:19 libmysqlclient_r.so.18.1.0 -> libmysqlclient.so.18.1.0
lrwxrwxrwx 1 root root 20 Nov 19 23:19 libmysqlclient.so -> libmysqlclient.so.20
lrwxrwxrwx 1 root root 24 Nov 19 23:19 libmysqlclient.so.18 -> libmysqlclient.so.18.1.0
-rwxr-xr-x 1 root root 9580608 Sep 13 17:07 libmysqlclient.so.18.1.0
lrwxrwxrwx 1 root root 24 Nov 19 23:18 libmysqlclient.so.20 -> libmysqlclient.so.20.3.7
-rwxr-xr-x 1 root root 9884704 Sep 13 17:15 libmysqlclient.so.20.3.7
-rw-r--r-- 1 root root 44102 Sep 13 17:13 libmysqlservices.a
drwxr-xr-x 4 root root 28 Nov 19 23:18 mecab
drwxr-xr-x. 3 root root 4096 Nov 19 23:19 plugin
The .so compiles and runs fine on the dev box, but copying it to a box without mysql-devel installed yields the following error:
cannot load module "zbx_mysql.so": libmysqlclient.so.20: cannot open shared object file: No such file or directory
I can only assume this means that the libmysqlclient.so.20.so isn't being bundled into my .so. I'm pretty much a novice here, so if anyone can advise it'd be greatly appreciated.
Shared libraries aren't "bundled", that's why they're shared. The machine you're trying to run on obviously misses the library. Libraries typically aren't in the "-dev" or "-devel" packages.
On your typical *nix system, you can have multiple versions of the same shared library installed, but normally only one development package. If you have the dev package for mysql-client 20 installed, the compiled code will link against that version. If you want your compiled code to link against mysql-client 18, install the older version of the development package.
If you need to be independent of the libraries installed on your target system, one possibility would be to link a static library instead.
Is there any link/documentation available around installing DC/OS on Google Compute Engine where instances are ubuntu 16.04 instances including the bootstrap node instead of CentOS 7?
Currently , the documents I find use Ansible and CentOS 7 on GCE as below.
https://dcos.io/docs/1.7/administration/installing/cloud/gce/
Short answer: Debian based distributions are currently (at least up to DC/OS 1.10) not supported.
Long answer: It's possible, but requires some extra steps.
DC/OS doesn't use any RedHat specific features. Most important differences could be solved by symlinks for few system binaries, as RedHat systems have different paths and systemd doesn't support $PATH variable in service definition. You'll need following:
sudo apt-get install libcurl3-nss ipset selinux-utils curl unzip bc
sudo ln -s /bin/mkdir /usr/bin/mkdir
sudo ln -s /bin/ln /usr/bin/ln
sudo ln -s /bin/tar /usr/bin/tar
sudo ln -s /bin/rm /usr/bin/rm
sudo ln -s /usr/sbin/useradd /usr/bin/useradd
sudo ln -s /bin/bash /usr/bin/bash
sudo ln -s /sbin/ipset /usr/sbin/ipset
Another requirements are:
systemd with version >=200
Docker >=1.6
Slightly outdated scripts from John Omernik, there's also puppet module (I'm the author). For further details see discussion on DC/OS Jira.
Next step is manual DC/OS compilation (it might sound scary, but actually it's very easy). C++ components (especially mesos-slave) are dependent on system libraries and they'd better be linked to proper libraries.
apt install python3-venv build-essential git
git clone https://github.com/dcos/dcos
./build_local.sh
Resulting "image" will be located in:
$HOME/dcos-artifacts/testing/`whoami`/dcos_generate_config.sh
You can copy it to your bootstrap server and extract:
bash dcos_generate_config.sh --genconf
after updating genconf/config.yaml you can start a container for serving the installation scripts:
docker run -d -p 9090:80 -v $PWD/genconf/serve:/usr/share/nginx/html:ro nginx
On a new node simply fetch the installation script:
rm -rf /tmp/dcos && mkdir /tmp/dcos && cd /tmp/dcos && curl -O http://bootstrap.example.com:9090/dcos_install.sh
bash dcos_install.sh slave
Unless you don't want to run packages from DC/OS Universe (like Elastic, Kafka, etc.) that depends on libmesos-bundle, you might be just fine. The bundle is fetched into each executor's directory, it includes numerous libraries, such as libmesos.so
...
-rwxr-xr-x 1 nobody nogroup 55077256 Jun 28 19:50 libmesos-1.4.0.so
-rwxr-xr-x 1 nobody nogroup 1487 Jun 28 19:50 libmesos.la
lrwxrwxrwx 1 nobody nogroup 17 Jun 28 19:50 libmesos.so -> libmesos-1.4.0.so
-rwxr-xr-x 1 nobody nogroup 398264 Jun 28 19:53 libpcre.so.1
-rwxr-xr-x 1 nobody nogroup 121296 Jun 28 19:53 libsasl2.so.3
-rwxr-xr-x 1 nobody nogroup 155744 Jun 28 19:53 libselinux.so.1
-rwxr-xr-x 1 nobody nogroup 454008 Jun 28 19:53 libssl.so.10
-rwxr-xr-x 1 nobody nogroup 999944 Jun 28 19:53 libstdc++.so.6
-rwxr-xr-x 1 nobody nogroup 79000 Jun 28 19:53 libsvn_delta-1.so.0
-rwxr-xr-x 1 nobody nogroup 1820208 Jun 28 19:53 libsvn_subr-1.so.0
-rwxr-xr-x 1 nobody nogroup 20040 Jun 28 19:53 libuuid.so.1
-rwxr-xr-x 1 nobody nogroup 90664 Jun 28 19:53 libz.so.1
drwxr-xr-x 3 nobody nogroup 4096 Jun 28 19:53 mesos
drwxr-xr-x 2 nobody nogroup 4096 Jun 28 19:37 pkgconfig
Some libraries might be compatible with your system, but the versions between CentOS and Debian might (and will) differ. You might encounter errors like:
libmesos-bundle/lib/libcurl.so.4: version `CURL_OPENSSL_3' not found (required by curl)
which will cause that all agent based health checks that use curl won't work, therefore most instances will refuse to start.
I have Perl installed on Linux RHEL:
perl -v
This is perl 5, version 12, subversion 2 (v5.12.2) built for x86_64-linux-thread-multi
When I run command to see Perl modules installed:
find `perl -e 'print "#INC"'` -name '*.pm' -print
I can see for DBI:
/mu/sdk/perl/5.12.2-gcc443-rhel5-64/lib/DBI/DBD/Metadata.pm
/mu/sdk/perl/5.12.2-gcc443-rhel5-64/lib/DBI/DBD/SqlEngine.pm
/mu/sdk/perl/5.12.2-gcc443-rhel5-64/lib/DBD/mysql/GetInfo.pm
/mu/sdk/perl/5.12.2-gcc443-rhel5-64/lib/DBD/mysql.pm
/mu/sdk/perl/5.12.2-gcc443-rhel5-64/lib/Bundle/DBD/mysql.pm
And MySQL installed and running:
Server version: 5.6.26 MySQL Community Server (GPL)
I have 2 lines of code in a Perl Script:
use DBI;
my $dbh = DBI->connect("DBI:mysql:database=test;host=localhost","root", "pass");
Error that I see is:
$ perl mysql.pl
install_driver(mysql) failed: Can't load '/mu/sdk/perl/5.12.2-gcc443-rhel5-64/lib/auto/DBD/mysql/mysql.so' for module DBD::mysql: libmysqlclient.so.15: cannot open shared object file: No such file or directory at /mu/apps/perl/5.12.2-gcc443-rhel5-64/lib/DynaLoader.pm line 200.
at (eval 3) line 3
Compilation failed in require at (eval 3) line 3.
Perhaps a required shared library or dll isn't installed where expected
at mysql.pl line 2
$
I verified with YUM:
$ sudo yum install perl-DBD-MySQL
Setting up Install Process
Package perl-DBD-MySQL-4.013-3.el6.x86_64 already installed and latest version
Nothing to do
$
This is the list of /usr/lib64/mysql :
$ pwd
/usr/lib64/mysql
$ ls -lrth
/usr/lib64/mysql
total 14M
-rwxr-xr-x 1 root root 8.6M Jul 14 17:47 libmysqlclient.so.18.1.0*
-rwxr-xr-x 1 root root 2.7M Jul 14 17:49 libmysqlclient.so.16.0.0*
-rwxr-xr-x 1 root root 2.7M Jul 14 17:49 libmysqlclient_r.so.16.0.0*
lrwxrwxrwx 1 root root 24 Jul 27 18:23 libmysqlclient.so.18 -> libmysqlclient.so.18.1.0*
lrwxrwxrwx 1 root root 24 Jul 27 18:23 libmysqlclient_r.so.18.1.0 -> libmysqlclient.so.18.1.0*
lrwxrwxrwx 1 root root 20 Jul 27 18:23 libmysqlclient_r.so.18 -> libmysqlclient.so.18*
lrwxrwxrwx 1 root root 24 Jul 27 18:23 libmysqlclient.so.16 -> libmysqlclient.so.16.0.0*
lrwxrwxrwx 1 root root 26 Jul 27 18:23 libmysqlclient_r.so.16 -> libmysqlclient_r.so.16.0.0*
drwxr-xr-x 3 root root 4.0K Jul 27 18:23 plugin/
$
What should I do next?
How to configure this problematic (libmysqlclient.so.15) or there is some other problem?
I have installed the following packages --
libevent-2.0.21
memcached-1.4.17
libmemcached-0.34
memcached_functions_mysql_1.1
All of the above have been installed successfully.
The output of the ldconfig -v command (the part where it shows that libmemcached libraries have been included) is as follows:
$ldconfig -v
/usr/local/libmemcached/lib:
libmemcached.so.3 -> libmemcached.so.3.0.0
libmemcachedprotocol.so.0 -> libmemcachedprotocol.so.0.0.0
libmemcachedutil.so.0 -> libmemcachedutil.so.0.0.0
But when I try to load UDF's into mysql using the install_functions.sql that is shipped with memcache_functions it throws the following error:
ERROR 1126 (HY000) at line 38: Can't open shared library 'libmemcached_functions_mysql.so' (errno: 0 libmemcached.so.3: cannot open shared object file: No such file or directory)
And the contents of the plugin directory are:
-rw-r--r-- 1 root root 6.1K Jan 21 13:49 adt_null.so
-rw-r--r-- 1 root root 11K Jan 21 13:49 auth.so
-rw-r--r-- 1 root root 6.0K Jan 21 13:49 auth_socket.so
-rw-r--r-- 1 root root 6.2K Jan 21 13:49 auth_test_plugin.so
-rw-r--r-- 1 root root 35K Jan 21 13:49 ha_example.so
-rw-r--r-- 1 root root 10K Jan 21 13:49 libdaemon_example.so
-rw-r--r-- 1 root root 361K Feb 13 02:47 libmemcached_functions_mysql.a
-rwxr-xr-x 1 root root 1.1K Feb 13 02:47 libmemcached_functions_mysql.la
-rwxr-xr-x 1 root root 167K Feb 13 02:47 libmemcached_functions_mysql.so
-rwxr-xr-x 1 root root 167K Feb 13 02:47 libmemcached_functions_mysql.so.0
-rwxr-xr-x 1 root root 167K Feb 13 02:47 libmemcached_functions_mysql.so.0.0.0
-rw-r--r-- 1 root root 11K Jan 21 13:49 mypluglib.so
-rw-r--r-- 1 root root 5.9K Jan 21 13:49 qa_auth_client.so
-rw-r--r-- 1 root root 11K Jan 21 13:49 qa_auth_interface.so
-rw-r--r-- 1 root root 6.0K Jan 21 13:49 qa_auth_server.so
-rw-r--r-- 1 root root 39K Jan 21 13:49 semisync_master.so
-rw-r--r-- 1 root root 15K Jan 21 13:49 semisync_slave.so
Installing memcached on Ubuntu 12.04.1 LTS (64 bit)
Python - 2.6.8
MySQL – 5.5.28
Zope – 2.12.19
1. Install memcached
apt-get install memcached
2. Install libmemcached -0 .34 (Version is very important. May or may not work with other versions).
Download it from https://launchpad.net/libmemcached/
tar xvf libmemcached-0.34.tar.gz
sudo ./configure --prefix=/usr/lib/libmemcached --with-memcached=/usr/bin/memcached
sudo make
sudo make install.
3. Install memcached_functions_mysql (Version 1.1 used at the time of the installation. (To create UDF’s that are invoked by triggers to manipulate the cache).
Downloading from https://launchpad.net/memcached-udfs
sudo ./configure --prefix=/usr/local/memcached_mysql --libdir=/usr/lib/mysql/plugin --with-mysql=/usr/bin/mysql_config --with-libmemcached=/usr/lib/libmemcached
sudo make
sudo make install
Navigate to the “sql” folder inside the memcached_mysql_functions directory.
mysql –u <username> -p < install_functions.sql
4. For Zope users
Install python-memcached-1.53
Download it from https://pypi.python.org/pypi/python-memcached/
Navigate to the extracted python-memcached directory.
/home/zope/zope/bin/python setup.py install
5. Edit the script containing which imports the modules.
allow_module('memcache')
from memcache import Client
allow_class(Client)
allow_module('memcache.Client.get')
allow_module('memcache.Client.set')
This is done so that memcache can be imported and used in your Restricted Python scripts.
If an external method is used to handle the above case, then the file does need to be updated with the above content.