Re: [lxc-users] The dark side of LXC
Did you mean 10.0.12? 10.0.1 is an alpha release https://mariadb.com/kb/en/mariadb/development/release-notes/release-notes-mariadb-100-series/mariadb-1001-release-notes/ Also, as a side note, if you have another program which uses libmysqlclient18 AND does version checking (e.g. php-5.3) AND using mariadb 10 to replace the distro's bundled mysql server, you might see sudden flood of warning messages (e.g. on php's error log) saying mismatch version between headers and client library version. There are ways to workaround this issue though, one of them is to NOT replace the distro's version of mysql client, and install mariadb separately from generic binary tarball. The other method is have the server and client on separate containers/servers. Just a heads-up in case anyone stumbles on this problem. -- Fajar On Mon, Jul 14, 2014 at 11:03 AM, CDR vene...@gmail.com wrote: I switched to MariaDB 10.0.1. It works fine. On Sun, Jul 13, 2014 at 10:30 PM, Fajar A. Nugraha l...@fajar.net wrote: Distros often change things to make it more suitable with their environment, or to fix bugs faster than upstream. For example, Ubuntu's version of mysql uses only upstart, which only tracks the pid of the msyqld process it started. I've also looked at Mariadb's[1] startup script, and it should handle service startup/shutdown correctly since it uses pid file. So you can do one of: - file a bug to oracle, and hope they fix it. - switch to mariadb - switch to a distro that properly fix the bug - fix the bug yourself (e.g. modify the startup script, based on one that works correctly) -- Fajar [1] https://mariadb.org/ On Sat, Jul 12, 2014 at 11:50 PM, CDR vene...@gmail.com wrote: I think it is a bug from Oracle, not my distro. They act like if the owned the server. On Sat, Jul 12, 2014 at 12:42 PM, Stéphane Graber stgra...@ubuntu.com wrote: On Sat, Jul 12, 2014 at 11:35:06AM -0400, CDR wrote: There is an unresolved issue that really goes to the core of the container model. Yesterday I installed Mysql latest on the host, and had to remove it one hour later. It turns out that since I have 20 containers each with a running version if mysql, the instance running on the host kills the other instances when you try to stop mysql, an similarly, the host version never starts because it thinks, wrongly, that it is already running. We need to make the host ignore the processes running on the containers, much like a real virtual machine, or else the host cannot really run any application whatsoever. That's the sign of a broken init script, init scripts should be tracking the pid of the spawned process and only kill that. Please file a bug against your distro, that's nothing to do with LXC. Philip Orleans ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users -- Stéphane Graber Ubuntu developer http://www.ubuntu.com ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users
Re: [lxc-users] The dark side of LXC
Distros often change things to make it more suitable with their environment, or to fix bugs faster than upstream. For example, Ubuntu's version of mysql uses only upstart, which only tracks the pid of the msyqld process it started. I've also looked at Mariadb's[1] startup script, and it should handle service startup/shutdown correctly since it uses pid file. So you can do one of: - file a bug to oracle, and hope they fix it. - switch to mariadb - switch to a distro that properly fix the bug - fix the bug yourself (e.g. modify the startup script, based on one that works correctly) -- Fajar [1] https://mariadb.org/ On Sat, Jul 12, 2014 at 11:50 PM, CDR vene...@gmail.com wrote: I think it is a bug from Oracle, not my distro. They act like if the owned the server. On Sat, Jul 12, 2014 at 12:42 PM, Stéphane Graber stgra...@ubuntu.com wrote: On Sat, Jul 12, 2014 at 11:35:06AM -0400, CDR wrote: There is an unresolved issue that really goes to the core of the container model. Yesterday I installed Mysql latest on the host, and had to remove it one hour later. It turns out that since I have 20 containers each with a running version if mysql, the instance running on the host kills the other instances when you try to stop mysql, an similarly, the host version never starts because it thinks, wrongly, that it is already running. We need to make the host ignore the processes running on the containers, much like a real virtual machine, or else the host cannot really run any application whatsoever. That's the sign of a broken init script, init scripts should be tracking the pid of the spawned process and only kill that. Please file a bug against your distro, that's nothing to do with LXC. Philip Orleans ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users -- Stéphane Graber Ubuntu developer http://www.ubuntu.com ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users
Re: [lxc-users] The dark side of LXC
On Sat, Jul 12, 2014 at 11:35:06AM -0400, CDR wrote: There is an unresolved issue that really goes to the core of the container model. Yesterday I installed Mysql latest on the host, and had to remove it one hour later. It turns out that since I have 20 containers each with a running version if mysql, the instance running on the host kills the other instances when you try to stop mysql, an similarly, the host version never starts because it thinks, wrongly, that it is already running. We need to make the host ignore the processes running on the containers, much like a real virtual machine, or else the host cannot really run any application whatsoever. That's the sign of a broken init script, init scripts should be tracking the pid of the spawned process and only kill that. Please file a bug against your distro, that's nothing to do with LXC. Philip Orleans ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users -- Stéphane Graber Ubuntu developer http://www.ubuntu.com signature.asc Description: Digital signature ___ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users