Re: Linux HA with RHEL and MySQL on System z
On Sat, Mar 13, 2010 at 09:18:09AM -0500, Alan Altmark wrote: On Saturday, 03/13/2010 at 06:49 EST, Frank Pani fpa...@ca.ibm.com wrote: 1) Because the database will be 200-300GB, we can use LVM and have it available to mount RW on either system. We need to be very careful no to mount RW at the same time; else we risk destroying the data. In cases where customers are using VMWARE or Xen on distributed, you ensure this by using PaceMaker to tell the hypervisor to issue a STONITH to totally kill the primary machine. This ensures, that the filesystem will be RW only on the backup now. However! There is currently no capability on z/VM to allow PaceMaker (or any other software that I know of) to do this. For quite a while now we have had the System Management API (SMAPI). Write a program on Linux that uses SMAPI to stop the Linux guest. Have PaceMaker execute said program. Such a program exists. It is called snipl and is available on developerWorks: http://www.ibm.com/developerworks/linux/linux390/snipl.html It's also included in some distributions. Also the Redbook Achieving High Availability on Linux for System z with Linux-HA Release 2 might be of interest: http://www.redbooks.ibm.com/abstracts/sg247711.html -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux HA with RHEL and MySQL on System z
On Sat, 2010-03-13 at 09:18 -0500, Alan Altmark wrote: On Saturday, 03/13/2010 at 06:49 EST, Frank Pani fpa...@ca.ibm.com wrote: 1) Because the database will be 200-300GB, we can use LVM and have it available to mount RW on either system. We need to be very careful not to mount RW at the same time; else we risk destroying the data. In cases where customers are using VMWARE or Xen on distributed, you ensure this by using PaceMaker to tell the hypervisor to issue a STONITH to totally kill the primary machine. This ensures, that the filesystem will be RW only on the backup now. However! There is currently no capability on z/VM to allow PaceMaker (or any other software that I know of) to do this. For quite a while now we have had the System Management API (SMAPI). Write a program on Linux that uses SMAPI to stop the Linux guest. Have PaceMaker execute said program. Linux provides the snipl command (incl. a STONITH plugin) to kill a system (LPAR or z/VM guest). It is available with SLES10 / SLES11 or can be downloaded from: http://www.ibm.com/developerworks/linux/linux390/snipl.html Ursula Braun Linux on System z development IBM Boeblingen -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: problem deinstalling an rpm
Just a sidenote.: never ever think about using --force unless you have a consistent valid working perfect backup AND every other option failed AND you know what you are doing. The last time I've tried it (couple of years ago), I ended up with a kernel panic on a server 80 miles from me in the middle of the night. Not funny, I assure... Mauro http://mauro.limeiratem.com - registered Linux User: 294521 Scripture is both history, and a love letter from God. On Wed, Mar 10, 2010 at 2:55 PM, Bruce Furber bfur...@comcast.net wrote: Try --force May the force be with you CHAPLIN, JAMES (CTR) james.chap...@associates.dhs.gov wrote: I am having a problem with a possible corrupted rpm. I tried to uninsta;; and rpm, and it tells be it is not installed to begin with, go to install, message is that is it installed. r...@zn001:(/root)#rpm -i Some_modulefiles.s390x.rpm package Some_modulefiles.s390x is already installed r...@zn001:(/root)#rpm -q Some_modulefiles.s390x.rpm package Some_modulefiles.s390x.rpm is not installed same message with rpm -e . . . . is not installed and quits. James Chaplin Systems Programmer, MVS, zVM zLinux Base Technologies, Inc Supporting the zSeries Platform Team -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
2010-03-12 Linux on System z kernel 2.6.33 related updates on developerWorks
Please refer to http://www.ibm.com/developerworks/linux/linux390/whatsnew.html for several 'Development stream' updates: * description of upstream kernel 2.6.33 features that were contributed by Linux on System z development * kernel 2.6.33 patches for kerntypes and kernel message catalog * s390-tools 1.8.4 delivers various usability enhancements and bug fixes * snIPL 2.1.7 with bugfixes * updated Linux on System z manuals, matching kernel 2.6.33 and other upstream packages: - Device Drivers, Features, and Commands - Using the Dump Tools - How to use FC-attached SCSI devices with Linux on System z - How to use Execute-in-Place Technology with Linux on z/VM - Kernel Messages * updated 'Development stream' Technical Details and Restrictions * end of message Mit freundlichen Grüßen / Kind regards Gerhard Hiller Systems Software Management IBM Systems Technology Group, Systems Software Development Phone: +49-7031-16-4388 IBM Deutschland Fax: +49-7031-16-3545 Schoenaicher Str. 220 E-Mail: ghil...@de.ibm.com 71032 Boeblingen Germany IBM Deutschland Research Development GmbH / Vorsitzender des Aufsichtsrats: Martin Jetter Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294 image/gifimage/gifimage/gifimage/gifimage/gifimage/gifimage/gif
Re: JAVA 1.6 BUG - Polling Loop
Thanks Rob, Will investigate further. Gerard -Original Message- From: Linux on 390 Port [mailto:linux-...@vm.marist.edu] On Behalf Of Rob van der Heij Sent: Sunday, March 14, 2010 4:28 PM To: LINUX-390@vm.marist.edu Subject: Re: JAVA 1.6 BUG - Polling Loop On Sun, Mar 14, 2010 at 4:32 PM, Shockley, Gerard C gsh...@bu.edu wrote: Some time ago it was asserted that there was an issue within the Java 1.6 code where an active timer loop was implemented in the JSDK for performance improvement. Thus causing an IDLE s390x system to appear to be using 99.X % CPU with HIGH CPUTIME for the java process. I realize several colleagues were aware (GT 6 mos ago) and I've substantially been able to stay clear of the 1.6 code, until today. New apps are driving this requirement. Can anyone provide additional insight? Who is working on it? My observations in this area were on JVM 1.5 that introduced a new optimizer that used polling to collect statistics for the profiler. According to my notes, JVM 1.5 SR4 introduced an unsupported and undocumented option to disarm it, and JVM 1.5 SR6 arranged to stop polling when there was no class active. So we should assume it's not present in your 1.6 JVM. But no guarantee that nothing else has been broken since... strace reveals active function call in futex() http://en.wikipedia.org/wiki/Futex This appears to be 100x worse then the old kernel timer pop event. There's also a kernel bug that rounds small delays to 0 which makes polling very painful. That would show by extremely high SIE intercepts (see ESACPUU) and high CP overhead. With the proper patches, the kernel will round the small delay up to 10 ms and reduce the CPU cost of polling (but it will still prevent the virtual machine dropping from queue when idle). If you're on a kernel 2.6.21 or later, you might want to pick up a copy of PowerTop which identifies the process that is polling (there's a s390x version at http://www.rvdheij.nl/linux/ ) You can match the PID against a java coredump and point fingers. Rob -- Rob van der Heij Velocity Software http://www.velocitysoftware.com/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: JAVA 1.6 BUG - Polling Loop
If you're on a kernel 2.6.21 or later, you might want to pick up a copy of PowerTop which identifies the process that is polling (there's a s390x version at http://www.rvdheij.nl/linux/ ) You can match the PID against a java coredump and point fingers. Do you have a sles10 version of that utility. Thanks, Gerard -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux HA with RHEL and MySQL on System z
This is all a good lead. I will have to let you folks know how it turns out. Thanks! - Frank Pani From: Ursula Braun ubr...@linux.vnet.ibm.com To: LINUX-390@vm.marist.edu Date: 03/15/2010 06:10 AM Subject:Re: Linux HA with RHEL and MySQL on System z On Sat, 2010-03-13 at 09:18 -0500, Alan Altmark wrote: On Saturday, 03/13/2010 at 06:49 EST, Frank Pani fpa...@ca.ibm.com wrote: 1) Because the database will be 200-300GB, we can use LVM and have it available to mount RW on either system. We need to be very careful not to mount RW at the same time; else we risk destroying the data. In cases where customers are using VMWARE or Xen on distributed, you ensure this by using PaceMaker to tell the hypervisor to issue a STONITH to totally kill the primary machine. This ensures, that the filesystem will be RW only on the backup now. However! There is currently no capability on z/VM to allow PaceMaker (or any other software that I know of) to do this. For quite a while now we have had the System Management API (SMAPI). Write a program on Linux that uses SMAPI to stop the Linux guest. Have PaceMaker execute said program. Linux provides the snipl command (incl. a STONITH plugin) to kill a system (LPAR or z/VM guest). It is available with SLES10 / SLES11 or can be downloaded from: http://www.ibm.com/developerworks/linux/linux390/snipl.html Ursula Braun Linux on System z development IBM Boeblingen -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: JAVA 1.6 BUG - Polling Loop
On Mon, Mar 15, 2010 at 7:53 PM, Shockley, Gerard C gsh...@bu.edu wrote: If you're on a kernel 2.6.21 or later, you might want to pick up a copy of PowerTop which identifies the process that is polling (there's a s390x version at http://www.rvdheij.nl/linux/ ) You can match the PID against a java coredump and point fingers. Do you have a sles10 version of that utility. Thanks, Gerard It requires function in the kernel that does not show before 2.6.21. Before that, I had to put a CP TRACE at interesting spots in the kernel to identify the offender. Rob -- Rob van der Heij Velocity Software http://www.velocitysoftware.com/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390