Re: What speed are my channels actually running at?
Great, That's what I hoped. Thanks.
Re: CP Query wildcards
If some folks can agree on what it should do, I'll do the writeup. I don't think it's going to be very high on the priority list, though, since the workaround is so easy. I think CP doesn't do wildcards because we didn't ask for it. We didn't a sk for it because we could throw together a QD EXEC to do what we needed, a lot quicker than wa iting for IBM to deliver a new release or two. Many CP commands do allow a range of (hex) device num bers dev1-dev2.
Re: CP Query wildcards
-Original Message- From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of David Boyes Sent: Wednesday, July 01, 2009 9:11 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: CP Query wildcards Wouldn't it be nice if you could 'ifconfig eth*'? You can in Plan 9. 8-) Why can't someone write an ifconfig replacement which accepts something like: ifconfig 'eth*' The shell would not expand this due to the ' marks. The ifconfig command process would therefore see the asterisk. It would then need to do whatever to inquire on the available eth* interfaces and report on them. This is almost like MS-DOS. The MS-DOS command prompt did not do any file name expansion, but passed the * on to the CP. Therefore, everything that wanted to accept wild cards had to implement the logic in themselves. I remember doing this. -- John McKown Systems Engineer IV IT Administrative Services Group HealthMarkets(r) 9151 Boulevard 26 * N. Richland Hills * TX 76010 (817) 255-3225 phone * (817)-961-6183 cell john.mck...@healthmarkets.com * www.HealthMarkets.com Confidentiality Notice: This e-mail message may contain confidential or proprietary information. If you are not the intended recipient, please contact the sender by reply e-mail and destroy all copies of the original message. HealthMarkets(r) is the brand name for products underwritten and issued by the insurance subsidiaries of HealthMarkets, Inc. -The Chesapeake Life Insurance Company(r), Mid-West National Life Insurance Company of TennesseeSM and The MEGA Life and Health Insurance Company.SM
Re: CP Query wildcards
I don't know what globbing is, but we have here an assembler routine that does generalized wildcard comparison between a field and a wildcard spec. The spec is built according to the rules used by the LISTFILE command of CMS: * = zero or more of any character, % = exactly one of any character. We have used this routine as the basis of several tools, and in one release of VM I even added it to the CMS nucleus and called it from the NAMEFIND module, allowing wildcards in the values supplied for tags. (Hmmm... it looks like that is no longer possible. If I am seeing what I think I am, NAMEFIND is now OCO. Sheesh... why???) I looked at the source for Q DASD volser, and at first glance it seems possible, and even not impractical, to add an exit point to HCPQDC in the code after label VOCONT (after the line with CLC VOLID,RDEVSERO) to call the the wildcard checking routine. If the real volser would be found to match the pattern supplied, the volser could then be accepted even though CP did not find a strict match. Unfortunately, I do not have time now to play with this, much as my fingers itch to try it. topic change On another topic that is connected to this thread: Just for neatness, when I write a pipe with a stage that is not always needed, rather than duplicate the pipe as was suggested by someone here, I prefer to put the optional stage in a variable: sortstage = '' ifSORTYN = 'Y' then sortstage = '| sort 21.6 ascending' 'pipe cp query dasd', '| locate /'VOLSER'/', sortstage, '| pad 35', '| chop 35', '| stem DASDX1.' (I am not commenting on the pipe itself, just on the duplication for the sort stage). Shimon Original message So ... to make 'q dasd' work the way David expresses, that function would need to call some kind of globbing subfunction. Same thing for 'ifconfig' in Linux and Unix. We need normalized namespace globbing, something which can be used by commands which don't necessarily do files, commands which are part of non-Unix stuff (like the CP Nuc). This stuff has been implemented in C time and again. Too bad CP doesn't have a C runtime instantiated. Oh ... wait ... It does! :-)
First OSA config.
I'm trying to figure out how to configure the OSA card on our z9 box. I'v e never worked with OSA's before so I have more questions than answers. AN Y help will be appreciated. The initial OSA config was done by a business partner. These are the IOCP def's : CHPID PATH=(10),TYPE=OSE,PCHID=110,PART=PROD CHPID PATH=(11),TYPE=OSC,PCHID=111,PART=PROD CNTLUNIT PATH=(10),UNIT=OSA,CUNUMBR=1460 IODEVICE ADDRESS=(1460,15),UNIT=OSA,UNITADD=00,CUNUMBR=1460 IODEVICE ADDRESS=(146F,1),UNIT=OSAD,UNITADD=FE,CUNUMBR=1460 * CNTLUNIT CUNUMBR=E020,PATH=(11),UNIT=OSC IODEVICE ADDRESS=(E020,08),CUNUMBR=(E020),UNIT=3270 We are using E020 E021 for our VM Prod VSE consoles. I've been trying to set up the CMS Users OSASF, OSAMAINT, OSADMIN1 2 3 using the OSA Implemntation redbook the OSA Customer guide. I have not done anything with user 4OSASF40, should I? I've gone thru chapter 4 (Setting up OSA/SF on z/VM) and I'm pretty sure I have everything defined as it says to. We're running z/VM 5.4 When I logon to OSADMIN1 and enter IOACMD, I get the menu back but when I select option 4 to get the config, I get : CMALLC_PRODUCT_SPECIFIC_ERROR: CMSIUCV CONNECT completed with return cod e 1011 The only explanation I've found to this message says I don't have TASF installed ??? Obviously I'm not doing something right! All ideas will be appreciated
Re: CP Query wildcards
Well, actually I use my CPE EXEC which is much more general: /* */ arg command 'PIPE CP' command '| CPE OUTPUT A' 'XEDIT CPE OUTPUT A' Then do any filtering I want in XEDIT. Brian On Wed, 1 Jul 2009 14:45:17 -0700, Schuh, Richard rsc...@visa.com wrote : What about devices attached to users? Regards, Richard Schuh -Original Message- From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of Brian Nielsen Sent: Wednesday, July 01, 2009 2:42 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: CP Query wildcards Try this QDASD EXEC: /* Query DASD by pattern */ arg dasd_info 'PIPE (ENDCHAR ?)', 'cp q dasd all', ' | SPLIT STR /, /', ' | l: LOCATE W3 /CP/', ' | SPECS W2 1 W5 NW / / N', ' | f: FANINANY', ' | LOCATE /'dasd_info'/', ' | qdasd output a', '?', 'l:', ' | SPECS W2 1 W3 NW / / N', ' | f:' 'X qdasd output a' Here are some sample calls, surrounded by quotes so extra blanks are obvious: 'QDASD RES ' -- will show volids that end in RES 'QDASD VM' -- will show volids that begin with VM 'QDASD PG'-- will show volids containing PG 'QDASD 54'-- will show volids or addresses containing 54= 'QDASD FF'-- same as above for FF, but also shows OFFLIN= E Brian Nielsen On Wed, 1 Jul 2009 12:39:05 -0400, Dean, David (I/S) david_d...@bcbst.com wrote: Can I have it? David Dean Information Systems *bcbstauthorized* From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of Wakser, David Sent: Wednesday, July 01, 2009 12:14 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: CP Query wildcards Because the command syntax requires a DASD pack name. I once wrote an EXEC to perform what you are attempting to do. David Wakser From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of Dean, David (I/S) Sent: Wednesday, July 01, 2009 11:59 AM To: IBMVM@LISTSERV.UARK.EDU Subject: CP Query wildcards Why won't my wildcard work? q dasd l53701 DASD C018 CP SYSTEM L53701 1 DASD C0CE CP SYSTEM L53701 1 DASD C0CF CP SYSTEM L53701 1 CP q dasd l53* DASD L53* was not found. Thanks David Dean Information Systems *bcbstauthorized* - = ===
Re: CP Query wildcards
On Thu, Jul 2, 2009 at 2:40 PM, McKown, Johnjmck...@healthmarkets.com wrote: This is almost like MS-DOS. The MS-DOS command prompt did not do any file name expansion, but passed the * on to the CP. Therefore, everything that wanted to accept wild cards had to implement the logic in themselves. I remember doing this. Puh. I found that the MS Opens Hell supports the autofill with TAB (dunno how Linux calls that) by taking it as any file that starts - this is even worse than the MS-DOS Command Prompt that just moved to the point where it could not decide anymore. PS Sloppy typing on SCP commands resulted in some unwanted files like u...@hostname.domain.com on my workstation. I am very happy to use them with the TAB key when typing in ssh or scp commands :-) Rob
Re: First OSA config.
Issue HELP MACRO CMSIUCV, then it will tell you to look for the return code in manual CP system Facilities (or something alike) for the error codes of IUCV CONNECT 2009/7/2 Frankovich, Bob bfrankov...@scspa.com I'm trying to figure out how to configure the OSA card on our z9 box. I've never worked with OSA's before so I have more questions than answers. ANY help will be appreciated. The initial OSA config was done by a business partner. These are the IOCP def's : CHPID PATH=(10),TYPE=OSE,PCHID=110,PART=PROD CHPID PATH=(11),TYPE=OSC,PCHID=111,PART=PROD CNTLUNIT PATH=(10),UNIT=OSA,CUNUMBR=1460 IODEVICE ADDRESS=(1460,15),UNIT=OSA,UNITADD=00,CUNUMBR=1460 IODEVICE ADDRESS=(146F,1),UNIT=OSAD,UNITADD=FE,CUNUMBR=1460 * CNTLUNIT CUNUMBR=E020,PATH=(11),UNIT=OSC IODEVICE ADDRESS=(E020,08),CUNUMBR=(E020),UNIT=3270 We are using E020 E021 for our VM Prod VSE consoles. I've been trying to set up the CMS Users OSASF, OSAMAINT, OSADMIN1 2 3 using the OSA Implemntation redbook the OSA Customer guide. I have not done anything with user 4OSASF40, should I? I've gone thru chapter 4 (Setting up OSA/SF on z/VM) and I'm pretty sure I have everything defined as it says to. We're running z/VM 5.4 When I logon to OSADMIN1 and enter IOACMD, I get the menu back but when I select option 4 to get the config, I get : CMALLC_PRODUCT_SPECIFIC_ERROR: CMSIUCV CONNECT completed with return code 1011 The only explanation I've found to this message says I don't have TASF installed ??? Obviously I'm not doing something right! All ideas will be appreciated -- Kris Buelens, IBM Belgium, VM customer support
Re: PAV and minidisks...
I don't have a full answer to your question. But I want to avoid a misconception: - mutipathing in z Architecture means a device can be reached by more than one path, most often this means more than one CHPID leads to the device, and each CHPID is connected to a different controlunit. - Without PAV: when a device in handling an IO, other IOs will be queued, for example by CP (reported by Pefkit). But also Linux, SFS, DB2VM, xxx know that classically a device can handle one one IO, and will queue other IOs (not reported by Perfkit). - PAV at the other hand makes it possible to have more than 1 I/O active on a single device. PAV is kind of a ly: a given device address can still have only one IO active; with PAV one assigns alternate device addresses to a single device. With PAV: when CP gets IO requests from different users for the same device, it will look for a free PAV address and may be able to launch it instead of queueing it. Linux -as far as I know- is also PAV aware, so it can launch more than one IO on condition that one gives it PAV addresses, otherwise it won't be able to exploit it. 2009/6/30 RPN01 nix.rob...@mayo.edu Before I put something huge together to test this, I thought I’d pass it by all the experts. Linux has the ability to multipath, and z/VM supports multipathing via PAV. There’s lots of documentation and studies showing that you can attach / dedicate the PAV addresses to a Linux LPAR or guest, and implement multipathing to DASD devices. This seems to be fairly clearly researched and understood. What I’m wondering about would be multiple links to the same minidisk (partial 3390, as opposed to a full volume) backed by a PAV multi-address environment sustained by z/VM. Would it help I/O throughput to have multiple MW minidisks set up in Linux as multipathing, if they were on a DASD with PAV enabled, and having several physical addresses? Are there any got’chas to this configuration? Any reason why it wouldn’t work? -- Robert P. Nix Mayo Foundation.~. RO-OE-5-55 200 First Street SW/V\ 507-284-0844 Rochester, MN 55905 /( )\ -^^-^^ In theory, theory and practice are the same, but in practice, theory and practice are different. -- Kris Buelens, IBM Belgium, VM customer support
Re: PAV and minidisks...
Your response verifies what I¹d thought was happening, but doesn¹t address the whole ³multiple writable minidisk² quandary. I¹m considering something like the following: USER LINUXGUEST MDISK 391 3390 1500 500 VOL001 MW LINK * 391 1391 MW LINK * 391 2391 MW Which would give me three virtual devices all pointing to the same minidisk within the Linux guest. First big question: Have I shot myself in the foot? Common z/VM wisdom says that multiple write enabled links to the same minidisk lead down a slippery slope to disaster. But would that be the case here? Second big question: Would PAV see the various I/O requests and assign them to separate PAV aliases, allowing for better thorughput to the device? I¹m thinking that, if I can get past the first question, then the second would be ³yes². One thing you didn¹t mention in your response is that, hopefully, many of the requests can be satisfied from cache, either via MDC or control unit caching, avoiding the actual I/O. The thing that PAV and the multiple minidisks would give you is the ability to get those I/Os started sooner than with a single path in Linux. -- Robert P. Nix Mayo Foundation.~. RO-OE-5-55 200 First Street SW/V\ 507-284-0844 Rochester, MN 55905 /( )\ -^^-^^ In theory, theory and practice are the same, but in practice, theory and practice are different. On 7/2/09 12:15 PM, Kris Buelens kris.buel...@gmail.com wrote: I don't have a full answer to your question. But I want to avoid a misconception: * mutipathing in z Architecture means a device can be reached by more than one path, most often this means more than one CHPID leads to the device, and each CHPID is connected to a different controlunit. * Without PAV: when a device in handling an IO, other IOs will be queued, for example by CP (reported by Pefkit). But also Linux, SFS, DB2VM, xxx know that classically a device can handle one one IO, and will queue other IOs (not reported by Perfkit). * PAV at the other hand makes it possible to have more than 1 I/O active on a single device. PAV is kind of a ly: a given device address can still have only one IO active; with PAV one assigns alternate device addresses to a single device. * With PAV: when CP gets IO requests from different users for the same device, it will look for a free PAV address and may be able to launch it instead of queueing it. Linux -as far as I know- is also PAV aware, so it can launch more than one IO on condition that one gives it PAV addresses, otherwise it won't be able to exploit it. 2009/6/30 RPN01 nix.rob...@mayo.edu Before I put something huge together to test this, I thought I¹d pass it by all the experts. Linux has the ability to multipath, and z/VM supports multipathing via PAV. There¹s lots of documentation and studies showing that you can attach / dedicate the PAV addresses to a Linux LPAR or guest, and implement multipathing to DASD devices. This seems to be fairly clearly researched and understood. What I¹m wondering about would be multiple links to the same minidisk (partial 3390, as opposed to a full volume) backed by a PAV multi-address environment sustained by z/VM. Would it help I/O throughput to have multiple MW minidisks set up in Linux as multipathing, if they were on a DASD with PAV enabled, and having several physical addresses? Are there any got¹chas to this configuration? Any reason why it wouldn¹t work?
Re: Linux install on a z/VM 1st level with an OSA defined as layer 2 network
Thanks, We were able to get past the FTP error. The issue was within the layer 2 VSWITCH definitions. We needed to have the VLAN and the NAT LAN statement set to the same LAN name ( 515 ). Now we are able to ping the VM machine on the layer 2 along with the linux guest when the first phase of the install has completed. Now we are running into another problem which seems to be with the Linux install script issue working on the layer 2 network. The first phase does runs, but the phase 2 part fails. Below are the console messages leading up to the failure and the VNC connection error. Enabling syn flood protection..done Disabling IP forwarding..done Disabling IPv6 forwarding..done Enabling IPv6 autoconfig..done Disabling IPv6 privacy..done ..done System Boot Control: The system has been set up System Boot Control: Running /etc/init.d/boot.local [1A..doneStarting syslog services..done Starting D-BUS daemon..done Starting HAL daemon..done Setting up network interfaces: lo loIP address: 127.0.0.1/8 Checking for network time protocol daemon (NTPD): ..unused Can't determine current runlevel [1A..doneWaiting for mandatory devices: eth-id-02:00:51:50:00:00 __NSC__ 19 18 17 16 15 14 13 12 11 10 9 8 7 6 4 3 2 1 0 eth-id-02:00:51:50:00:00No interface found [1A..failedSetting up service network . . . . . . . . . . . . . . . ...failed ip_tables: (C) 2000-2006 Netfilter Core Team MORE... SANDBOX Jul 1 16:46:08 linux kernel: ip_tables: (C) 2000-2006 Netfilter Core Team *** *** Please return to your X-Server screen to finish installation *** starting VNC server... A log file will be written to: /var/log/YaST2/vncserver.log ... *** *** You can connect to , display :1 now with vncviewer *** Or use a Java capable browser on http://:5801/ *** (When YaST2 is finished, close your VNC viewer and return to this window.) *** Fatal Error occurred, process stopped *** - Commandline available at Alt-F2 - Further information written to: /var/log/YaST2/y2start.log Russell Gendreau Time Customer Service, Inc. 813-554-2064 -Original Message- From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of Tom Duerbusch Sent: Friday, June 26, 2009 4:22 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: Linux install on a z/VM 1st level with an OSA defined as layer 2 network No real trick to it. First, make sure you have everything setup properly on the FTP server. On the same VM as your FTP server, log on to CMS, and try a FTP. This makes sure your FTP server is setup properly. Then, on the same VM as you are doing the Linux install, logon to CMS and try the FTP. If successful, you have a good server and good communications. As installed, the FTP server is not configured for what you want to do. The following is a snipit of the changes I have to make: Install vsFTPd ftp server Must be done to be an install server. When done, don't need winscp for FTP transfers, but it will still work. yast Network Services Network Services (xinetd) Enable ftp: edit vsftpd: continue (to install vsftpd) ok (installed successfully) (enable) service is active accept finish quit joe /etc/vsftpd.conf comment out: Listen=YES eliminates 500 OOPS: count not bind listening IPv4 sockets uncomment: local_enable=yes write_enable =yes save and exit Tom Duerbusch THD Consulting russell.gendr...@custserv.com 6/26/2009 2:44 PM Greetings We have successfully setup an OSA as a layer 2 network with the defined vswitch and vlan. We are having an issue with the build of a Linux guest defined for a layer 2 network. While the Linux build is starting it goes out to another Linux guest on another z/VM LPAR to download the binaries. *** Error while accessing the FTP server: Failed to connect to FTP server *** Could not find the SUSE Linux Enterprise Server 10 Installation Source. Activating manual setup program. Linuxrc v2.0.79 (Kernel 2.6.16.60-0.21-default) Has anyone ever successfully built a Linux guest under z/VM on a layer 2 network? Are there any tricks to get the installs to work? The theory is that with the layer 2 network in place we will be able to move Linux guest from one z/VM to another and potentially move from one box to another box as long as the OSA's on them are connected to the same layer 2 network and that the OSA's are defined the same so that the Linux guest can retain their IP address's no matter which box they are running. Supposedly for load balancing and when maintenance is required on the z/VM systems there will be minimal
Re: CP Query wildcards
Thanks to all for the help. Please, Sir, I want some more. Oliver PIPE cp q dasd | zone w5 wildcard /VMA*/ | cons FPLSCB027E Entry point WILDCARD not found FPLMSG003I ... Issued from stage 2 of pipeline 1 FPLMSG001I ... Running zone w5 wildcard /VMA*/ Ready(-0027); T=0.01/0.01 14:20:12 David Dean Information Systems *bcbstauthorized* -Original Message- From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of Rob van der Heij Sent: Wednesday, July 01, 2009 6:02 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: CP Query wildcards or PIPE cp q dasd | zone w5 wildcard /VMA*/ | cons;-) - Please see the following link for the BlueCross BlueShield of Tennessee E-mail disclaimer: http://www.bcbst.com/email_disclaimer.shtm
Re: CP Query wildcards
David, As noted in the sample QDASD EXEC that I posted, the 'WILDCARD' stage is provided only by the CMS Pipelines Runtime Library (aka: Pipe RTL) available at: http://vm.marist.edu/~pipeline/ It is my belief that most 'advanced' z/VM customer sites that do not have some form of restriction on loading public domain software run using Pipes RTL as their default. Pipes RTL has many improved, and many altogether new, stages compared with the older IBM z/VM Endicott Lab version distributed with z/VM. Pipes RTL is backward and upward compatible with Endicott Pipes. Pipes RTL is authored and supported (on an ad hoc, but very good basis) by John Hartmann of IBM Denmark, the author of CMS Pipelines. Read the PIPELINES NEWS files at that URL for descriptions of some of the newer enhancements. When there, be sure to load down the installation instructions, as well as the PICKPIPE EXEC (which makes it easier to swap back and forth between the Pipes RTL and Endicott distributions). Once you are convinced that Pipes RTL is the one true, enlightened path, you may want to go back to the Endicott Pipes on occasion when testing something, PICKPIPE gives you an easy means of doing so. Mike Walter Hewitt Associates Any opinions expressed herein are mine alone and do not necessarily represent the opinions or policies of Hewitt Associates. Dean, David (I/S) david_d...@bcbst.com Sent by: The IBM z/VM Operating System IBMVM@LISTSERV.UARK.EDU 07/02/2009 01:24 PM Please respond to The IBM z/VM Operating System IBMVM@LISTSERV.UARK.EDU To IBMVM@LISTSERV.UARK.EDU cc Subject Re: CP Query wildcards Thanks to all for the help. Please, Sir, I want some more. Oliver PIPE cp q dasd | zone w5 wildcard /VMA*/ | cons FPLSCB027E Entry point WILDCARD not found FPLMSG003I ... Issued from stage 2 of pipeline 1 FPLMSG001I ... Running zone w5 wildcard /VMA*/ Ready(-0027); T=0.01/0.01 14:20:12 David Dean Information Systems *bcbstauthorized* -Original Message- From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On Behalf Of Rob van der Heij Sent: Wednesday, July 01, 2009 6:02 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: CP Query wildcards or PIPE cp q dasd | zone w5 wildcard /VMA*/ | cons;-) - Please see the following link for the BlueCross BlueShield of Tennessee E-mail disclaimer: http://www.bcbst.com/email_disclaimer.shtm The information contained in this e-mail and any accompanying documents may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient of this message, or if this message has been addressed to you in error, please immediately alert the sender by reply e-mail and then delete this message, including any attachments. Any dissemination, distribution or other use of the contents of this message by anyone other than the intended recipient is strictly prohibited. All messages sent to and from this e-mail address may be monitored as permitted by applicable law and regulations to ensure compliance with our internal policies and to protect our business. E-mails are not secure and cannot be guaranteed to be error free as they can be intercepted, amended, lost or destroyed, or contain viruses. You are deemed to have accepted these risks if you communicate with us by e-mail.
Re: PAV and minidisks...
The simple MW approach is surely wrong, it will not create a PAV environment: Linux will think it has 3 different devices, accidents will happen. MDC will avoid the IO; Control Unit cache hit is still IO as concerned for the z Series, but here PAV would help. AFAIK, PAV will not help if the concurrent IOs are not satisfied mostly from the control unit cache: the real disk can only handle one IO anyhow. I never implemented PAV (my former customer didn't have PAV enabled for the VM disks: it wouldn't help with DB2 nor SFS, and that's what they used heavily). You need to use the MDISK's MINIOPT directory record to tell CP to create a PAV group; keyword PAVALIAS. This way Linux will recognize all addresses as a PAV group. My guess: MDISK 391 3390 1500 500 VOL001 M (I removed the W) MINIOPT PAVALIAS 1391 2391 would create 391 as base and 1391 plus 2391 as PAV alias addresses. 2009/7/2 RPN01 nix.rob...@mayo.edu Your response verifies what I’d thought was happening, but doesn’t address the whole “multiple writable minidisk” quandary. I’m considering something like the following: USER LINUXGUEST MDISK 391 3390 1500 500 VOL001 MW LINK * 391 1391 MW LINK * 391 2391 MW Which would give me three virtual devices all pointing to the same minidisk within the Linux guest. First big question: Have I shot myself in the foot? Common z/VM wisdom says that multiple write enabled links to the same minidisk lead down a slippery slope to disaster. But would that be the case here? Second big question: Would PAV see the various I/O requests and assign them to separate PAV aliases, allowing for better thorughput to the device? I’m thinking that, if I can get past the first question, then the second would be “yes”. One thing you didn’t mention in your response is that, hopefully, many of the requests can be satisfied from cache, either via MDC or control unit caching, avoiding the actual I/O. The thing that PAV and the multiple minidisks would give you is the ability to get those I/Os started sooner than with a single path in Linux. -- Robert P. Nix Mayo Foundation.~. RO-OE-5-55 200 First Street SW/V\ 507-284-0844 Rochester, MN 55905 /( )\ -^^-^^ In theory, theory and practice are the same, but in practice, theory and practice are different. On 7/2/09 12:15 PM, Kris Buelens kris.buel...@gmail.com wrote: I don't have a full answer to your question. But I want to avoid a misconception: - mutipathing in z Architecture means a device can be reached by more than one path, most often this means more than one CHPID leads to the device, and each CHPID is connected to a different controlunit. - Without PAV: when a device in handling an IO, other IOs will be queued, for example by CP (reported by Pefkit). But also Linux, SFS, DB2VM, xxx know that classically a device can handle one one IO, and will queue other IOs (not reported by Perfkit). - PAV at the other hand makes it possible to have more than 1 I/O active on a single device. PAV is kind of a ly: a given device address can still have only one IO active; with PAV one assigns alternate device addresses to a single device. - With PAV: when CP gets IO requests from different users for the same device, it will look for a free PAV address and may be able to launch it instead of queueing it. Linux -as far as I know- is also PAV aware, so it can launch more than one IO on condition that one gives it PAV addresses, otherwise it won't be able to exploit it. 2009/6/30 RPN01 nix.rob...@mayo.edu Before I put something huge together to test this, I thought I’d pass it by all the experts. Linux has the ability to multipath, and z/VM supports multipathing via PAV. There’s lots of documentation and studies showing that you can attach / dedicate the PAV addresses to a Linux LPAR or guest, and implement multipathing to DASD devices. This seems to be fairly clearly researched and understood. What I’m wondering about would be multiple links to the same minidisk (partial 3390, as opposed to a full volume) backed by a PAV multi-address environment sustained by z/VM. Would it help I/O throughput to have multiple MW minidisks set up in Linux as multipathing, if they were on a DASD with PAV enabled, and having several physical addresses? Are there any got’chas to this configuration? Any reason why it wouldn’t work? -- Kris Buelens, IBM Belgium, VM customer support
Re: PAV and minidisks...
Thank you Kris; I think that¹s the piece I was looking for. :) -- Robert P. Nix Mayo Foundation.~. RO-OE-5-55 200 First Street SW/V\ 507-284-0844 Rochester, MN 55905 /( )\ -^^-^^ In theory, theory and practice are the same, but in practice, theory and practice are different. On 7/2/09 3:47 PM, Kris Buelens kris.buel...@gmail.com wrote: The simple MW approach is surely wrong, it will not create a PAV environment: Linux will think it has 3 different devices, accidents will happen. MDC will avoid the IO; Control Unit cache hit is still IO as concerned for the z Series, but here PAV would help. AFAIK, PAV will not help if the concurrent IOs are not satisfied mostly from the control unit cache: the real disk can only handle one IO anyhow. I never implemented PAV (my former customer didn't have PAV enabled for the VM disks: it wouldn't help with DB2 nor SFS, and that's what they used heavily). You need to use the MDISK's MINIOPT directory record to tell CP to create a PAV group; keyword PAVALIAS. This way Linux will recognize all addresses as a PAV group. My guess: MDISK 391 3390 1500 500 VOL001 M (I removed the W) MINIOPT PAVALIAS 1391 2391 would create 391 as base and 1391 plus 2391 as PAV alias addresses. 2009/7/2 RPN01 nix.rob...@mayo.edu Your response verifies what I¹d thought was happening, but doesn¹t address the whole ³multiple writable minidisk² quandary. I¹m considering something like the following: USER LINUXGUEST MDISK 391 3390 1500 500 VOL001 MW LINK * 391 1391 MW LINK * 391 2391 MW Which would give me three virtual devices all pointing to the same minidisk within the Linux guest. First big question: Have I shot myself in the foot? Common z/VM wisdom says that multiple write enabled links to the same minidisk lead down a slippery slope to disaster. But would that be the case here? Second big question: Would PAV see the various I/O requests and assign them to separate PAV aliases, allowing for better thorughput to the device? I¹m thinking that, if I can get past the first question, then the second would be ³yes². One thing you didn¹t mention in your response is that, hopefully, many of the requests can be satisfied from cache, either via MDC or control unit caching, avoiding the actual I/O. The thing that PAV and the multiple minidisks would give you is the ability to get those I/Os started sooner than with a single path in Linux.
Re: Linux install on a z/VM 1st level with an OSA defined as layer 2 network
On 7/2/2009 at 1:00 PM, russell.gendr...@custserv.com wrote: -snip- [1A..doneWaiting for mandatory devices: eth-id-02:00:51:50:00:00 __NSC__ 19 18 17 16 15 14 13 12 11 10 9 8 7 6 4 3 2 1 0 eth-id-02:00:51:50:00:00No interface found [1A..failedSetting up service network . . . . . . . . . . . I don't recall this one, but it looks as though the configuration file wasn't written out by the installer to use Layer 2, so it's trying to bring it up as a Layer 3 NIC. Try re-starting the installation, and after the DASD are activated, SSH in a second time to look at the config files in the /etc/sysconfig/hardware/ and /etc/sysconfig/network/ directories to make sure things look right. Mark Post
Mark Workman/IS/Shelter is out of the office.
I will be out of the office starting 07/02/2009 and will not return until 07/13/2009. I will respond to your message when I return. This e-mail is intended only for its addressee and may contain information that is privileged, confidential, or otherwise protected from disclosure. If you have received this communication in error, please notify us immediately by e-mailing postmas...@shelterinsurance.com; then delete the original message.
Re: PAV and minidisks...
On Thu, Jul 2, 2009 at 10:47 PM, Kris Buelenskris.buel...@gmail.com wrote: The simple MW approach is surely wrong, it will not create a PAV environment: Linux will think it has 3 different devices, accidents will happen. You're wrong. A very simple experiment could have shown you that: lxrob1:~ # head -n 1 /sys/bus/ccw/devices/0.0.04*2/uid == /sys/bus/ccw/devices/0.0.0402/uid == IBM.6800033025.0120.28.08bb0bda == /sys/bus/ccw/devices/0.0.0412/uid == IBM.6800033025.0120.28.08bb0bda lxrob1:~ # vmcp q md 402 loc TargetID Tdev OwnerID Odev Dtype Vol-ID Rdev StartLoc Size ROBLX1 0402 ROBLX1 0402 3390 VM4V05 0148 2235800 lxrob1:~ # vmcp q md 412 loc TargetID Tdev OwnerID Odev Dtype Vol-ID Rdev StartLoc Size ROBLX1 0412 ROBLX1 0402 3390 VM4V05 0148 2235800 Note: The casual observer may spot the beginning and end of the extent in the VM appendix to the device identification. MDC will avoid the IO; Control Unit cache hit is still IO as concerned for the z Series, but here PAV would help. AFAIK, PAV will not help if the concurrent IOs are not satisfied mostly from the control unit cache: the real disk can only handle one IO anyhow. Not entirely. The current DASD subsystems under the covers also do various magic that may enable the back-end to spread a logical ECKD volume over multiple physical drives and thus have more than one I/O active at the same time on a single logical volume. It is fairly easy to demonstrate whether it works. One reason I did not (yet) react on Bob's questions is that it is rather complicated to determine which workload would benefit from it, if any. I stopped working on the draft response twice because it got way too long to post ;-) In general, I don't put energy in letting a single virtual machine monopolize any subsystem. On z/VM you normally have sufficient others that share the resources. Performance measurements that try to drive a single virtual machine as fast as possible have little value in my world. Rob -- Rob van der Heij Velocity Software http://www.velocitysoftware.com/
Re: First OSA config.
The first question, Bob, is Why are you using OSASF at all?? OSASF is not needed/used by the ICC port of the OSA. If you change your definition of the other port (PCHPID 110) to OSD and run in QDIO mode you won't need OSASF. You can have zVM and multiple guests all use/share that port witth no OSASF. zVM could have ports 1460-1461, guest 1 could have 1462-1463, guest 3 could have 1464-1465, etc Of course you will have to change your zvm and guest TCP/IP definitions to use QDIO mode, but that is generally very simple and this avoids the complications of OSASF. Of course, the zVM folks here will also probably point out that you could just define a VSWITCH and connecct the port to that and then connect the zVM IP stack and all the guersts to that VSWITCH. This all does assume that your guests will support QDIO mode. Old systems may present a problem. Mike - Original Message - From: Frankovich, Bob bfrankov...@scspa.com To: IBMVM@LISTSERV.UARK.EDU Sent: Thursday, July 02, 2009 9:35 AM Subject: First OSA config. I'm trying to figure out how to configure the OSA card on our z9 box. I'v= e never worked with OSA's before so I have more questions than answers. AN= Y help will be appreciated. The initial OSA config was done by a business = partner. These are the IOCP def's : CHPID PATH=(10),TYPE=OSE,PCHID=110,PART=PROD CHPID PATH=(11),TYPE=OSC,PCHID=111,PART=PROD CNTLUNIT PATH=(10),UNIT=OSA,CUNUMBR=1460 IODEVICE ADDRESS=(1460,15),UNIT=OSA,UNITADD=00,CUNUMBR=1460 IODEVICE ADDRESS=(146F,1),UNIT=OSAD,UNITADD=FE,CUNUMBR=1460 * CNTLUNIT CUNUMBR=E020,PATH=(11),UNIT=OSC IODEVICE ADDRESS=(E020,08),CUNUMBR=(E020),UNIT=3270 We are using E020 E021 for our VM Prod VSE consoles. I've been trying to set up the CMS Users OSASF, OSAMAINT, OSADMIN1 = 2 3 using the OSA Implemntation redbook the OSA Customer guide. I have= not done anything with user 4OSASF40, should I? I've gone thru chapter 4 (Setting up OSA/SF on z/VM) and I'm pretty sure I have everything defined= as it says to. We're running z/VM 5.4 When I logon to OSADMIN1 and enter IOACMD, I get the menu back but when I= select option 4 to get the config, I get : CMALLC_PRODUCT_SPECIFIC_ERROR: CMSIUCV CONNECT completed with return cod= e 1011 The only explanation I've found to this message says I don't have TASF = installed ??? Obviously I'm not doing something right! All ideas will be appreciated =
Re: PAV and minidisks...
On Tue, Jun 30, 2009 at 8:52 PM, RPN01nix.rob...@mayo.edu wrote: Before I put something huge together to test this, I thought I’d pass it by all the experts. Too bad they must all have responded to you off-list.. I would have loved to see a concise answer. But failing that, let me give it a try. Linux has the ability to multipath, and z/VM supports multipathing via PAV. There’s lots of documentation and studies showing that you can attach / dedicate the PAV addresses to a Linux LPAR or guest, and implement multipathing to DASD devices. This seems to be fairly clearly researched and understood. There's no such thing as enough performance research ;-) What I’m wondering about would be multiple links to the same minidisk (partial 3390, as opposed to a full volume) backed by a PAV multi-address environment sustained by z/VM. Would it help I/O throughput to have multiple MW minidisks set up in Linux as multipathing, if they were on a DASD with PAV enabled, and having several physical addresses? Are there any got’chas to this configuration? Any reason why it wouldn’t work? Many years ago I demonstrated that I could fool Linux multi path support by giving it multiple write links to the same mini disk (not sure it would still work after the changes in Linux and z/VM). I referred to that as virtual PAV until that term was hijacked by z/VM Development for something else ;-) The reason for doing this is that you would open up access to data in MDC while waiting for another I/O on the same mini disk (that was not satisfied by MDC). The diagnose I/O works very well in that aspect because it provides a fast path for data that does not require a real I/O. Unfortunately, the current Linux PAV support does not work with the diagnose driver. That's where I lost interest in it and stopped bugging VM development about fixing something in MDC. If we were interested in increasing the peak bandwidth for a single guest, then I think some of this could make sense. The PAV alias devices are a limited resource, so it is a waste to dedicate them to virtual machines that are only utilized 5% of the time (just like it is a waste to own a Porsche if you only drive it 5% of the time - darn, wrong analogy). When you leave the PAV aliases to z/VM, they can be shared among the virtual machines. With static PAV, the alias devices would be shared among the virtual machines sharing the base volume. With HyperPAV that would even work for pseudo full pack mini disks. Disk I/O performance for Linux on z/VM is not a trivial research subject. I once counted 18 layers of caching, queueing, blocking, striping, reordering, etc. You can't treat that like a black box if you want to predict what lab experiments mean for your business workload. Since then, I found a few more, and your scenarios add a some as well. Rob -- Rob van der Heij Velocity Software http://www.velocitysoftware.com/
Running RACFCONV on 5.4.0 (was 5.3 Tips)
Hi list, i found this posting while searching for information regarding RACFCONV. Just one question: How can i run RACFCONV on my 5.3.0 using the 5.4.0 templates? I found no hints in the documentation ... On z/OS i am just using a different STEPLIB. Any help appreciated. Tobias Doerkes. On Thu, 16 Aug 2007 16:06:23 -0400, Jim Bohnsack jab...@cornell.edu wro te: I'm bringing up z/VM 5.3 and came across a couple of items that aren't exactly plainly stated in the SDO doc. Both have to do with RACF. Also both are related to or caused by the fact that I really don't like to do a big bang type of upgrade. I build a new system on a 2nd level guest and then like to move it piece meal to my test lpar. In this case I am putting 5.3 in on the test system running z/VM 4.4. The first item or assist is one that I got from Colin Allison. I knew about the new database templates for RACF and, I guess, in recalling the trauma of having to convert the RACF DB to the structured format 10-15 years ago when you were making an irreversible change to the DB, I was nervous about doing it. Colin told me, however, that you can do the RACFCONV to add the new templates to the DB and then still use that DB running the older RACF. I did that and it worked just fine. I've been running the RACF that came with 4.4 with the 5.3 templates for a month or so with no problems. The only hint in the RACF prog. dir. is that if you are using a shared DB, you must convert the templates from the system with the highest level of RACF. The 2nd item that caught me is that today I decided to put CP 5.3 and the new RACF on the test lpar. RACF woudn't run. I had ipled with NOAUTOLOG and then logged onto RACFVM using the directory pw in order to switch to the new 490 and 305 disks. RACSTART would end with the rather crytpic message RACF is not defined to the Z/VM system. The problem is that the SYSTEM CONFIG file didn't have the new character string in it containing 5VMRAC30 that was inserted with the ENABLE command when I installed on the 2nd level system. It only had the old ENABLE statement containing the PRODID of 5767002P. I added the line containing 5VMRAC30 and it worked just fine. I just thought I'd pass on these tips for anyone who likes to upgrade a system in a little more granular manner than all at once. Jim Jim Bohnsack Cornell University (607) 255-1760 jab...@cornell.edu =
Re: Running RACFCONV on 5.4.0 (was 5.3 Tips)
On Friday, 07/03/2009 at 12:04 EDT, Tobias Doerkes tdoer...@hotmail.com wrote: Just one question: How can i run RACFCONV on my 5.3.0 using the 5.4.0 templates? I found no hints in the documentation ... You don't. You run RACFCONV on your 5.4 system against the 5.3 database to bring it up to 5.4. (Even though you continue to use the db with a 5.3 system.) Alan Altmark z/VM Development IBM Endicott