Re: Linux Slowdown
But at the same time, understanding what the monitor is telling me is absolutely complex. It's a lot of data to make sense of, especially how it coorelates together. I can get a sense of it, but I'm not sure that if something was really wrong I would be able to pinpoint the issue. -Original Message- From: Post, Mark K [mailto:[EMAIL PROTECTED] Sent: Thursday, December 02, 2004 7:53 PM To: [EMAIL PROTECTED] Subject: Re: Linux Slowdown Barton, You're absolutely right. Without data, all anyone can do is guess. I suspect the problem a lot of people are facing is that they don't (yet) have any sort of budget for a lot of the things they're trying out. I agree that anyone that has enough money to bring z/VM in house, but doesn't get a performance monitor to go along with it is just asking for trouble. But, some people don't yet have enough backing from management to do that. Too many people keep hearing that Linux is free (instead of Free), and they don't understand why it gets so expensive. Of course, doing things the right way has never been cheap, so I always wonder why they're surprised. But, Dilbert explains a lot of that, too. Mark Post -Original Message- From: Linux on 390 Port [mailto:[EMAIL PROTECTED] Behalf Of Barton Robinson Sent: Thursday, December 02, 2004 4:27 PM To: [EMAIL PROTECTED] Subject: Re: Linux Slowdown I really DO hate to bring this up, but It really does solve a lot of mysteries to have a performance monitor that collects your linux and VM data. Right now, i'm looking at some SAP data, Linux on z/VM, a big linux server logs off and the master processor utilization sky rockets, then the i/o from all of the other servers stops for a minute. now i know why service was degraded, and have enough information to understand the problem. Without data, how could i have a clue of which one of 100 common problems caused my problem this time? -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] 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 [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux Slowdown
On Thu, 2 Dec 2004 20:52:31 -0500, Post, Mark K <[EMAIL PROTECTED]> wrote: > some people don't yet have enough backing from management to do that. Too > many people keep hearing that Linux is free (instead of Free), and they > don't understand why it gets so expensive. Of course, doing things the Sure... Want to know how much time I wasted last month trying to look into a performance issue with a customer where I did not have the proper tools to do so? That time was probably worth a year's contract for the right tools. But sometimes you cannot get it done the way you like it. It is just so sad to see a proof of concept fail just because they did not get the right tools and people to fix the default tuning parameters. Rob -- Rob van der Heij rvdheij @ gmail.com -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux Slowdown
I too deal with this all the time. Many Pof C's consist of clients dumping PC server workload exactly as is into linux virtual machines. Not going to be kind to a "Z" box. When I ask for "Z" performance data or PC and SUN server data I usually get a blank stare. One client that is having major performance problems on "Z" linux is too busy to turn monitor on! Let alone have a performance tool. It's not that hard to instrument. And you want a silver bullet? 5 minute answer? What? One trend that really concerns me is the move to "P" series linux. Watch out: if "Z" P of C's are done improperly we will lose many opportunities to get linux on VM shops up running and stable - with acceptable performance and costs. Server data usually consists of a 1 page sheet listing the CPU speeds and memory size of the boxes. That's configuration data. When I finally prodded a client into a weeks worth of data from 8 servers the results were enlightening. End of the day it's nothing new, business as usual really. David Kreuter Rob van der Heij wrote: On Thu, 2 Dec 2004 20:52:31 -0500, Post, Mark K <[EMAIL PROTECTED]> wrote: some people don't yet have enough backing from management to do that. Too many people keep hearing that Linux is free (instead of Free), and they don't understand why it gets so expensive. Of course, doing things the Sure... Want to know how much time I wasted last month trying to look into a performance issue with a customer where I did not have the proper tools to do so? That time was probably worth a year's contract for the right tools. But sometimes you cannot get it done the way you like it. It is just so sad to see a proof of concept fail just because they did not get the right tools and people to fix the default tuning parameters. Rob -- Rob van der Heij rvdheij @ gmail.com -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] 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 [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux Slowdown
On Fri, 03 Dec 2004 10:37:12 -0500, David Kreuter <[EMAIL PROTECTED]> wrote: > linux. Watch out: if "Z" P of C's are done improperly we will lose many > opportunities to get linux on VM shops up running and stable - with > acceptable performance and costs. We're just in the front lines with zSeries. When I talk to my cow-orker who does Linux on VMware on xSeries, he is also concerned with systems management solutions that require 6 agents in Java running on each server... -- Rob van der Heij rvdheij @ gmail.com -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
2004-12-03 Recommended Linux on zSeries code drop to developerWorks
Please see the "What's new" page at: http://www10.software.ibm.com/developerworks/opensource/linux390/whatsnew.shtml Change summary: > "June 2003 stream": - Kernel 2.4.21: Recommended kernel bug fixes and retrofitted lcs driver from "April 2004 stream" kernel 2.6.5 Happy downloading! * end of message Mit freundlichem Gruß / Kind regards, Gerhard Hiller eServer Software Management, D4357 IBM Development Lab, Boeblingen/Germany Phone ext. +49-(0)7031 - 16 - 4388 Internet: [EMAIL PROTECTED] -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux Slowdown
I deal with this also. The problem I have is that I can't justify the cost of the tools when the PC solution works, and is actually less money when you add the cost of all the other stuff on the mainframe. -Original Message- From: Linux on 390 Port [mailto:[EMAIL PROTECTED] Behalf Of David Kreuter Sent: Friday, December 03, 2004 7:37 AM To: [EMAIL PROTECTED] Subject: Re: Linux Slowdown I too deal with this all the time. Many Pof C's consist of clients dumping PC server workload exactly as is into linux virtual machines. Not going to be kind to a "Z" box. When I ask for "Z" performance data or PC and SUN server data I usually get a blank stare. One client that is having major performance problems on "Z" linux is too busy to turn monitor on! Let alone have a performance tool. It's not that hard to instrument. And you want a silver bullet? 5 minute answer? What? One trend that really concerns me is the move to "P" series linux. Watch out: if "Z" P of C's are done improperly we will lose many opportunities to get linux on VM shops up running and stable - with acceptable performance and costs. Server data usually consists of a 1 page sheet listing the CPU speeds and memory size of the boxes. That's configuration data. When I finally prodded a client into a weeks worth of data from 8 servers the results were enlightening. End of the day it's nothing new, business as usual really. David Kreuter Rob van der Heij wrote: >On Thu, 2 Dec 2004 20:52:31 -0500, Post, Mark K <[EMAIL PROTECTED]> wrote: > > > >>some people don't yet have enough backing from management to do that. Too >>many people keep hearing that Linux is free (instead of Free), and they >>don't understand why it gets so expensive. Of course, doing things the >> >> > >Sure... Want to know how much time I wasted last month trying to look >into a performance issue with a customer where I did not have the >proper tools to do so? That time was probably worth a year's contract >for the right tools. But sometimes you cannot get it done the way you >like it. > >It is just so sad to see a proof of concept fail just because they did >not get the right tools and people to fix the default tuning >parameters. > >Rob >-- >Rob van der Heij rvdheij @ gmail.com > >-- >For LINUX-390 subscribe / signoff / archive access instructions, >send email to [EMAIL PROTECTED] 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 [EMAIL PROTECTED] 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 [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
IBM IHS apache based HTTP server - Verbose trace?
Does anyone here know how to make the IBM IHS 2.0 wbe server for linux do the equivalent of a '-VV' trace on the z/os HTTP server? -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: IBM IHS apache based HTTP server - Verbose trace?
On Fri, 2004-12-03 at 10:56, James Melin wrote: > Does anyone here know how to make the IBM IHS 2.0 wbe server for linux do > the equivalent of a '-VV' trace on the z/os HTTP server? > If you set "LogLevel debug" in httpd.conf, is the output still to vague? I've had some success performing a 'strace' on the httpd process. Off the top of my head, you'll need to use the -f and -F strace flags. Hope this helps. -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
DCSS updates in Bitkeeper
Hi Folks, for those who build Distributions or want to try new things first: today Linus merged much of our patches into the Bitkeeper tree, including new development for virtual server integration: - the DCSS block device driver now has a new parameter: as kernel parameter "dcssblk.segments=SEGONE(local),SEGTWO,SEGTHREE" will load SEGONE nonshared, and SEGTWO,SEGTHREE shared. as module parameter "segments=..." (same) - the extmem code that loads DCSSes has been rewritten and improved: better error messages (including VM diag condition code when diag fails) better usability when switching between shared/nonshared support for segments that mix EW and EN parts (for Rob van der Heij style swap segments): only EW or EN chunks allowed all chunks need to be contiguous (no gaps in the segment) maximum 6 chunks - and last but not least there is support for #cp def store config that enables using DCSSes in large-memory guests [has been there for a while already] All changes are supposed to be integrated in 2.6.10. With these changes, it will then be very handy to use DCSS as ultra-fast swap device: create a large dcss, for example from 256m to 1960m: #cp defseg SWAPPING 1-1 EW 10001-7a7ff EN #cp def store 2g #cp saveseg swapping #cp def store 256m have dcssblk built-in (not as module), and add the kernel parameter "dcssblk.segments=SWAPPING" run mkswap: mkswap /dev/dcssblk0 save swap segment: echo "1" >/sys/devices/dcssblk/SWAPPING/save add this line to /etc/fstab: /dev/dcssblk0 swap swap defaults 0 0 type swapon -a, and look at /proc/swaps to verify that everything worked fine Note that the current xip2 patch on linuxvm.org cannot be applied against 2.6.10 anymore, it needs to be adapted to the new (fixed) extmem interface with kind regards Carsten Otte -- omnis enim res, quae dando non deficit, dum habetur et non datur, nondum habetur, quomodo habenda est -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: DCSS updates in Bitkeeper
On Friday, 12/03/2004 at 06:13 CET, Carsten Otte <[EMAIL PROTECTED]> wrote: [snip] > > have dcssblk built-in (not as module), and add the kernel parameter > "dcssblk.segments=SWAPPING" > > > run mkswap: mkswap /dev/dcssblk0 > save swap segment: echo "1" >/sys/devices/dcssblk/SWAPPING/save > add this line to /etc/fstab: /dev/dcssblk0 swap swap defaults 0 0 > type swapon -a, and look at /proc/swaps to verify that everything worked > fine [insert] And then shutdown this Linux and bring up your class G exploiters. :-) Alan Altmark z/VM Development IBM Endicott -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: DCSS updates in Bitkeeper
On Dec 3, 2004, at 11:13 AM, Carsten Otte wrote: Hi Folks, for those who build Distributions or want to try new things first: today Linus merged much of our patches into the Bitkeeper tree, YAY! Adam -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Permission denied on Samba 3.0.9-17 download.
I'm getting permission denied when trying to download Samba 3.0.9-17 packages from sernet.de. URL: http://ftp.sernet.de/pub/samba/sles8-s390/ Can whomever it is that maintains this please open up the permissions on the files? Thanks *Brandon Darbro -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Effect of LIMITSOFT vs NOLIMIT?
Our current workload consists of a dozen Linux Oracle servers in a 1-cpu LPAR. CPU usage stays around 50% with very occasional spikes to 100%. I feel we have good resource allocation for these servers via SHARE REL. Default MAX is NOLIMIT and my question is whether LIMITSOFT would be of any benefit. Documentation I have read so far indicate that CPU will be limited unless unused resources are available, which seems to be the same as NOLIMIT, where other servers are getting their allocated minimums satisfied first. I'm sure its more involved - can anyone expand on this? Ray Mrohs Energy Information Administration U.S. Department of Energy -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Effect of LIMITSOFT vs NOLIMIT?
If NOLIMIT is set then it's possible for the share leftover when everyone has had their SHARE satisfied can be consumed by a single virtual machine. Using LIMITSOFT means this leftover will be shared according to the value set. Imagine an apple pie. After everyone has had 1 slice and there's some left over, then with NOLIMIT a greedy individual can gobble the whole lot. However, with a soft limit they have to be nice and share it with the other hungry souls (if there are still hungry people). -Original Message- Our current workload consists of a dozen Linux Oracle servers in a 1-cpu LPAR. CPU usage stays around 50% with very occasional spikes to 100%. I feel we have good resource allocation for these servers via SHARE REL. Default MAX is NOLIMIT and my question is whether LIMITSOFT would be of any benefit. Documentation I have read so far indicate that CPU will be limited unless unused resources are available, which seems to be the same as NOLIMIT, where other servers are getting their allocated minimums satisfied first. I'm sure its more involved - can anyone expand on this? -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Slack again
Any one seen this before using a lcs device? Pinging the gateway PING 65.212.159.1 (65.212.159.1): 56 data bytes lcs_fix_multicast_list failed to add multicast entry e001 multicast address table possibly full. --- 65.212.159.1 ping statistics --- 3 packets transmitted, 0 packets received, 100% packet loss -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Permission denied on Samba 3.0.9-17 download.
> Can whomever it is that maintains this please open up the > permissions on the files? Brandon, I've let Volker L know... I seriously doubt he's on this list. In the meantime if you are very anxious, I can send you a samba3-3.0.9-17.src.rpm. I can also send you s390 compiled rpms, but cannot guarantee they are the same as SerNet because I built them before the sernet s390 rpms were available. Let me know offlist... I'd want to check to make sure I didn't leave any of my crap laying around in the spec first. ~ Daniel -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: DCSS updates in Bitkeeper
On Fri, 3 Dec 2004 12:43:01 -0500, Alan Altmark <[EMAIL PROTECTED]> wrote: > And then shutdown this Linux and bring up your class G exploiters. :-) Would you be surprised to hear some people prefer to run a simple EXEC in a MAINT userid that creates the segment? ;-) Rob -- Rob van der Heij rvdheij @ gmail.com -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Slack again
Neale, I did a quick grep of the mailing list archives, and got a hit. According to Andy Cathrow on 2 Apr 2002 14:17:24, "The lcs module appears to load normally - apart from the multicast error (that the documentation says to ignore)". So, I would say that the message given is not the cause of any problems getting to the gateway. Mark Post -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Sent: Friday, December 03, 2004 10:54 AM To: [EMAIL PROTECTED] Subject: Slack again I don't believe this is a Slack specific problem but have you seen this before? Pinging the gateway PING 65.212.159.1 (65.212.159.1): 56 data bytes lcs_fix_multicast_list failed to add multicast entry e001 multicast address table possibly full. --- 65.212.159.1 ping statistics --- 3 packets transmitted, 0 packets received, 100% packet loss -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Open LDAP: TLS problems - forwarded mail
I am posting this on behalf of my colleague. We need to install Samba and join that with Active Directory, which requires Kerberos and Open LDAP. We have a problem installing TLS certificate. Anyone travelled this road before? TIA. __ Ranga Nathan / CSG Systems Programmer - Specialist; Technical Services; BAX Global Inc. Irvine-California Tel: 714-442-7591 Fax: 714-442-2840 - Forwarded by Ranga Nathan/AMERICA/BAX on 12/03/2004 05:20 PM - David Damon / CSG Operations, Senior Systems Integration Analyst; IT - Technical Services; Irvine-California 12/03/2004 05:18 PM To: Ranga Nathan/AMERICA/[EMAIL PROTECTED] cc: Subject:TLS problems Ranga, Here is what you can post to IBM and your news groups: OS: SuSE Linux Enterprise Server 9 Platform: IBM mainframe (390?) OpenLDAP version 2.2.6 Problem: I am trying to use TLS certificates and keys to secure communications with the LDAP server and client. When I implement the certificates and keys OpenLDAP crashes. The line in the log files is; Dec 3 11:43:48 zlintest1 slapd[10694]: main: TLS init def ctx failed: -1. I know it is the TLS entries in the slapd.conf file because I comment the entries out and the LDAP server comes up and stays up. The certificates and keys were generated using OpenSSL. It may be the way I generated the certificates and keys or how I answered the questions during generation. Regards, Senior Systems Integration Analyst Office: (714) 442-7562 FAX:(714) 442-2845 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390