In fact it was easy, it is right here at Velocity:
http://www.velocitysoftware.com/VMSTEAL.PDF
Van: Harder, Pieter
Verzonden: donderdag 21 januari 2021 17:41
Aan: Linux on 390 Port
Onderwerp: Re: Best stealing explanation for z/Linux under z/VM
I remember
I remember vaguely that long ago there was a paper by Rob vd Heij on steal
time. You will have to search for it. No idea if it is still available on the
list.
best regards,
Pieter Harder
Van: Linux on 390 Port namens Victor Echavarry
Verzonden:
NetBIOS/LAN Manager systems do you still
have? They’re about the only thing I can think of that would still care about
the old way. Anything post-Win9x with service packs should be able to do the
Kerberos stuff.
> On Mar 31, 2019, at 6:15 PM, Harder, Pieter
> wrote:
>
> Not if you
Not if you AD is still running in NTLM mode.
Van: Linux on 390 Port namens David Boyes
Verzonden: zondag 31 maart 2019 20:43
Aan: LINUX-390@VM.MARIST.EDU
Onderwerp: Re: zLinux authentication on windows AD LDAP
> Is it technically possible to
Van: Harder, Pieter
Verzonden: dinsdag 22 januari 2019 23:40
Aan: Linux on 390 Port
Onderwerp: Re: Examples of tape access
Long ago, in a different galaxy I had Ficon tapes library-mounted for Linux
guest through a companion CMS server (hello WAKEP!!) that did the DFSMS stuff.
Crude
Long ago, in a different galaxy I had Ficon tapes library-mounted for Linux
guest through a companion CMS server (hello WAKEP!!) that did the DFSMS stuff.
Crude, but it worked.
best regards
Pieter Harder
(still daily comparing VMWare ESX with z/VM and finding it lacking...)
In the past IBM has been extremely reluctant to outright own a Linux distro.
Likely for fear of alienating the Linux people by corporate behaviour. Why now?
Is this a preventive strike to prevent somebody else from owning Red Hat?
Microsoft seems to be on the prowl to buy open source stuff and
Just to give an extreme example from another time, another galaxy. I had a
z9-BC (I told you, long ago) with TSM Linux VM (Tivoli Storage Manager, the
backup thingy) that moved data for replication purposes from a TS3500 to a
TS3310 with six tape pairs. Reached almost 1 GB/s with very little
I think this practice of allocating 8G/8G has come from the virtualised Windows
world. There you need at least 8GB of memory to run the Windows server
independent from what you run on it. A 10k service module or an 8G (virtual)
SAP system. Yes, smallish SAP systems are quite good at living in
ged database which may have HUGE impact on
performance.
03-02-2016 01:28:09 "Harder, Pieter" wrote:
>In fact the disk system didn't change with our migration. The VMWare farm
was
>run against the same IBM DS8000 the z used. Most of the DS8000 space was
used
>for open syst
EDU> namens Philipp Kern
<pk...@debian.org>
Verzonden: dinsdag 2 februari 2016 22:24
Aan: LINUX-390@VM.MARIST.EDU
Onderwerp: Re: TCO comparison Intel vs EC12 IFLs
On 2016-02-02 14:54, Harder, Pieter wrote:
> I think there is another factor involved. On x86 you usually run
> da
Hi,
I think there is another factor involved. On x86 you usually run database
engines etc with very large caches to avoid going to disk. On z this is much
less of a penalty because of the powerful I/O subsystem. I have some experience
with this as we moved our SAP systems from z to x86 as part
David,
I remember from long ago that CMS NETDATA (not the TSO version) was documented
in the CMS manuals. I short search reveals it to be in the 620 CMS Macros and
Functions Reference.
best regards,
Pieter Harder
Van: Linux on 390 Port
Hi Berry,
Yes, this policy of non-support of Oracle on virtual systems applies to VMware
as well. That is why we still run our main Oracle on Sparc iron. Political
trouble ahead ;-)
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
The whole balance is different on a z. There often just doing the I/O
(especially when you can do lots of them at the *same* time, thus overlapping
them in time) is faster than rummaging around in bufferspace. My experience in
the past has been that on z database engines need an order of
Hi,
I am not a current S/390 customer, but when I was I spent quite some time
together with Rob vd Heij analysing VSwitch performance and the workings of the
virtualised instructions (none of them published by IBM) that go with them. At
that time (z/VM 5.3 ish) those instructions needed to
Mauro,
If you are running on z/VM there is an equivalent CMS based tool called
SCSIDISC. It is in the CMS maintenance tree and takes some renaming/copying of
components to activate. I found SCSIDISC easier to use than the equivalent
Linux based tools (if and when available).
Best regards,
Pieter, I've installed SCSIDISC here, but it could't find anything, so we
turned to Linux again.
You need to attach the FCP device(s) you have attached to the Linux guest to
the CMS running SCSIDISC. So:
1. shutdown and logoff owning Linux, freeing devices
2. attach same devices to CMS
3. run
Just a remark. TSM does *not* have a tape requirement. You can have a disk only
installation just fine. Do your first level backups to what your standard disk
for Linux uses. Then migrate to some cheap outside storage and have that
mirrored to a secondary location. My experience is that the
SAVESEG MONDCSS?
Van: Linux on 390 Port [LINUX-390@VM.MARIST.EDU] namens Dean, David (I/S)
[david_d...@bcbst.com]
Verzonden: maandag 6 juni 2011 20:53
Aan: LINUX-390@VM.MARIST.EDU
Onderwerp: MONDCSS after IPL
How do you make MONDCSS permanent? We IPL'd,
Can you describe in more detail what you exactly did?
I hope you are talking about manually changing a monolithic source directory
and not parts of a segmented directory on DIRMAINT disks? If the latter that
won't work and there is a fair chance you clobbered your directory. In that
case I hope
Yes, you can (at least for Suse I have done so). You illogically have to follow
the z/OS tree and Linux is linked in as a z/OS subproduct.
Van: Linux on 390 Port [linux-...@vm.marist.edu] namens Bauer, Bobby (NIH/CIT)
[E] [baue...@mail.nih.gov]
Just a guess, maybe you don't have VM64604 on your 540 system? And it is
probably integrated in 610. That one could change the native vlan without any
sound. I was hit by that one when I applied it.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
It is is not Intel, then it does not exist. How many Linux people (general
population) consider Linux on the z, i, or p? Unless you came up in the z, i,
or p world, those processors simply do not exist. computer is identically
equal to Intel processor.
Then why are you talking about Linux? You
Hi,
Wouldn't 'UseDNS no' in /etc/ssh/sshd_config do it?
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Agblad Tore
Verzonden: dinsdag 3 augustus 2010
The CPU's in the user directory are virtual ones, not related to real CPU's in
any way.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Sam Bass
Not the webcast, reading the Redbooks.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Marcy Cortes
Verzonden: donderdag 22 juli 2010 17:37
Aan:
Shouldn't that be LAYER2=1 in your conf?
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Bauer, Bobby
(NIH/CIT) [E]
Verzonden: maandag 14 juni 2010
Hi Ruddy,
I am running TSM V6.1 on a z9 with 2 IFL's. The production TSM runs with a
Vsize of 5GB. This one is moving a backup load of ~ 8TB a day and is rather
maxed out. Most of the LPAR is running SAP and that takes up most of the real
memory of 23GB with 1GB expanded. Another tip: do
You added packs. Nervous twitch: mkinitrd and zipl?
I think you can do the filsystem stuff with yast as well, by increasing the lv.
Van: Linux on 390 Port [linux-...@vm.marist.edu] namens LJ Mace
[ljmace1...@yahoo.com]
Verzonden: woensdag 2 juni 2010
I don't know about z10 (don't have one), but we run with two OSA 1000BaseT as a
failover pair shared between two z/OS and one z/VM/IFL lpars. The outside is
one /24 network and we gave out individual host addresses and use a VSWITCH
with two RDEVs on the z/VM side as well. The OSAs keep track
I don't know a thing about the Cisco switch, but I have 3 LPARs, 1 z/VM and
2 z/OS, sharing the same OSA Express-3 port. Since they are on the same
port they are obviously on the same subnet.
That is not necessarily true if you take vlan tagging into account. One
physical port may then
Are you sure you're doing the right thing there? It sounds like you've
started a database instance rather than DB2 Connect.
Eh Rob, DB2 Connect fat client contains a database instance.
You are thinking thin client.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel
Hi Billy,
My rule-of-very-big-thumb: one third of the vendor specified amount for x86.
My experience being with SAP and TSM. Both end up in the same area with us.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Folks,
We currently have our FCP attached storage to our DS8300 running
thru SVC. No NPIV. We are trying to set up now to be Native DS8300 using
NPIV. We have a lun attached (actually 4 on on image and 1 on another )
which seem to be working but our storage admin is NOT seeing the images
Yes, I had the same situation running release 3 code. The NPIV WWPNs were
logged, or they wouldn't have been usable. But the DS8300 admin UI (and CLI,
for that matter) showed them as not logged in. Move to release 4 solved that
for us, as promised (I had previously opened a pmh about this)
I suspect the main reason is that a big part of NPIV is implemented by zoning
in the switch. So without switch it is rather difficult to control who whacks
what. Not a very good idea in a production environment.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 /
pvscan?
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Ron Foster at
Baldor-IS
Verzonden: woensdag 24 februari 2010 15:15
Aan: LINUX-390@VM.MARIST.EDU
lvdisplay -m /dev/VG name/LV name
will display the physical volumes and extents for a logical volume.
Pvscan will tell what physical volumes belongs to what volume group, if any.
Lvscan will tell what logical volumes are defined.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel
Seems some operational hiccup if you got the configs right. What does lsqeth
tell you about the state of the nics?
Best regards,
Pieter Harder
Van: Linux on 390 Port [linux-...@vm.marist.edu] namens van Sleeuwen, Berry
[berry.vansleeu...@atosorigin.com]
Never seen it. I am running both SAP 640 and 701 on Linux SAP005
2.6.16.60-0.54.5-default.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Ron Foster
This is a lun on my installation:
0x40124000
The lun number in the Storage Admin is 1200
So that conforms exactly to x'40ab40cd'
If you can't see it, it is most likely zoning in the SAN switch. Have the
person who is responsible for that look into i.
Best regards,
Pieter Harder
ON z/VM 540 @ RSU0902:
SCSIDISC LOG A
INFO::START @ 15 Dec 2009 18:53:55
INFO::No of attached FCP Channels found=1
INFO::Choosen FCP Sub-Channel Device(s)=ALL DEVICES
INFO::FCP SUB-CHANNEL 401F Initialized
INFO::For Sub Channel 401F No of WWPNs found=1
INFO::For 401F Choosen
I think there is a more z/VM related issue: master-only work. One lpar
obviously has one master. By the time you get to the large lpars contemplated
for this thread contention for the master may become a problem. But then I have
no real knowledge of this. Probably more something for Barton and
When that happens you are either:
1. Not using NPIV, and thus suffering the consequences.
2. Or should immediately take your storage admins to the woodshed
for giving out unwarranted access rights.
Regarding 1, Rob's statement was correct because NPIV cannot be used when
point-to-point.
I
Hi Craig and Ron,
I am the customer that Rob's case study was based on. I had the exact same
problem (with 1 GbE OSA's) that you are having. Your problem is indeed the
VSWITCH. Kick it and use native bonding in Linux. If you need more than one
guest, buy more OSA's, they are way cheaper than
For the finer points of performance measurement I defer to Rob. This is a
highlevel coarse view:
With a VSWITCH with LACP involved and using two full IFL's with 4 OSA GbE I
could barely exceed 100 MB/s with no other activity at all. With the same 2
IFL's and 4 OSA dedicated and bonded within
...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
Van: Craig Collins [mailto:grizl...@gmail.com]
Verzonden: maandag 26 oktober 2009 20:58
Aan: Harder, Pieter
Onderwerp: OSA-Express3 10 Gb, Vswitch, and SLES10 Linux Throughput
Hi Pieter,
Thanks for your response. We've never tried connecting them directly
Do you have statistics for packets on each interface in the port group?
When I had the VSWITCH setup I was looking closely into this. While not
perfectly balanced due to our client distribution, there was reasonable traffic
going on on all interfaces. And it was the processor power available.
A sadly realistic checkpoint, though: You are unlikely to EVER get an
appreciable percentage of wire speed out of the 10G interfaces using TSM no
matter what you do; the protocol is half duplex in a lot of places which
kills the interface data pipelining.
If you can get it to exceed 15-20% of
Hi,
I had a look at aroud the same time. Dropped it in favorr of Velocity's ESALPS.
More function, less overhead. And never looked back.
Best regards,
Pieter Harder
Van: Linux on 390 Port [linux-...@vm.marist.edu] namens Ron Foster at Baldor-IS
Does anyone know if there is a supported version of the lin_tape Linux
device driver for SUSE LINUX 10 SP2?
The file lin_tape_359X_zSeries.ReadMe on
ftp.software.ibm.com/storage/devdrvr/Linux
only shows support up to SP1, kernel 2.6.16.46-0.12.
Yes, go one step deeper:
There will be plenty of us reminding you on this post when you found
something happened to the order of your disks in the virtual machine
configuration and the mkswap wiped out your root file system or some
other relevant data...
That shouldn't happen if you are using udev to address them
I agree that the speed of CPU is not the problem anymore. But do you want CPU
intensive load? Afaik the CPU in mainframe, even z10, is still more expensive
than an intel. So if you'd have CPU intensive loads, would it be still cost
effective to run on z? Do we still need to stay away from these
Hi Berry,
You did provide /etc/sysconfig/hardware for the second nic? If not the
/etc/sysconfig/network doesn't kick in as the hardware is not picked up.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux
-...@vm.marist.edu] On Behalf Of Harder,
Pieter
Sent: woensdag 17 juni 2009 13:29
To: LINUX-390@VM.MARIST.EDU
Subject: Re: How to activate 2nd nic
Hi Berry,
You did provide /etc/sysconfig/hardware for the second nic? If not the
/etc/sysconfig/network doesn't kick in as the hardware is not picked up
.
Regards, Berry.
-Original Message-
From: Linux on 390 Port [mailto:linux-...@vm.marist.edu] On Behalf Of Harder,
Pieter
Sent: woensdag 17 juni 2009 14:04
To: LINUX-390@VM.MARIST.EDU
Subject: Re: How to activate 2nd nic
Hi Berry,
After providing the hardware definition you need to either
I think it is not a layer2/3 issue. When that is wrong the nic will get an
error on activation, but the devices will come online. This sounds more like
some hangup in udev for the basic device.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
Hi,
If you didn't touch the config then simply forcing and xautologging dirmaint
will probably fix it.
I you deleted the definition of minidisk 123 from the dirmaint directory entry,
or the definition that may point to (eg maint 123) things get more complicated.
If so come back to the
Hi Paul,
So we'll keep checking and also thanks for the VSS info we had looked at
that in the past and for some reason it was dropped to we will review
that one again too.
One pitfall to think about I ran into with VSS the hard way is that VSS
operates according to the 'same box' principle.
1. We found that when you add OSAs to a Vswitch and setup link
aggregation that the OSAs is used only on that Vswitch.
That is by definition. Setting up layer2 LACP requires exclusivity for the OSA
chpid.
2. When you setup port channeling in the Real Switch you must be
reading
With Linux on z/VM I do like the idea that reduced
number of CPUs may improve chances of the virtual machine drop from
queue nicely. But recent SAP releases have stuff that makes even the
virtual 1-way stay in-queue, so there it does not help anymore.
And now for the really stupid question: does
Hi Jan,
How and where do we get that WWPN
Our storage guys do not understand our question
Storage guys that do not know what wwpn's are? Get new storage guys ;-))
Seriously, to start with basics:
- is your switch a Fibre channel switch and not a Ficon-only one?
- do they have fibre ports on the
Then we have to add the WWPN to to device
We are not sure what WWPN to use
On the switch or on the SAN itself
And how do we get to the WWPN to display (In other words where do we
find it)
If you are running on z/VM there is a tool called SCSIDISC. It is on the CP
object disk. With that you scan
Hello Martha,
You don't seem t have gotten many replies, so let me try.
However, I'm planning on making use of this on a z9 for our production
Linux guests who need access to large data volumes (i.e. our Oracle servers,
etc.).
I am not using edev but native FCP/NPIV. I have looked at edev in
Hi Lee,
Is anyone running a TSM server (not client) on Linux on Z? Any observations?
Yes, we are. Using a DS8300 for disk storage, a TS3500/LTO3 for primary tape
pools and a TS3310/LOT4 for copy pools and a 4x GbE LACP OSA bundle. It is a
completely automated environment with the two tape
Hi John,
We are considering Linux on z9, and I noticed you mentioning the overhead of
VSWITCH LACP - which we are planning to use. Do you have some detailed
information on this overhead? Is it affected by SET PORT GROUP INTERVAL ?
Any information would be gratefully received:-)
The problem
Have you checked the system vdisk limit? The help for HCPLNM091E states this is
a failure of either the user or system limit. And if you in fact exceeded the
system limit make sure you have the paging space to increase the system limit
or this could get very painfull indeed.
Best regards,
We've figured out how to setup a SAN and make it available to one of our
zLinux guest. The Ficon adapter was attached to the linux guest first with an
ATTACH command and now with a DEDICATE statement in user direct. That says to
me I can't use that Ficon adapter for any other guest that I may
I'm wondering if something like this would be a reasonable option for
people with less requirements than RACF/VM. It could even read the
directory source to get all passwords in, but maybe the separation of
the two is even a good thing. It could free VM development from some
of the dual-path
Nelson wrote:
We are setting up an SLES 10 SP2 Linux environment running under z/VM
5.4. We have not determined what
performance monitoring tool to use yet.
What tools are other shops using for performance monitoring? Thanks.
I have both ESALPS and VM PTK. I prefer ESALPS for Linux
How do then zone them so that the guest can see them
You don't need to have them zoned to define them within Linux. You only need to
have them zoned in the switch to use them.
The storage people say that they have to see the wwwpn before they can
zone it
That is not true. But they have to do
That would be my desire as well, but IBM says it is what it is for security
reasons.
I agree with them. Paranoia maybe, but better safe than sorry.
just a way to grant * or grant all. The way it is now is such a pain.
Exactly why I suggested 'grant public' in a previous post. But your
Hi Marcy,
If you are using the lin_tape driver and udev you can use static naming by
using /dev/lin_tape/by-id/wwpn-IBMtape. I am not quite sure about the old
IBMtape driver, but if you are using that one it is about time to move to
lin_tape anyway.
Best regards,
Pieter Harder
Does anybody know is it possible to create vswitch which will be
unaware to access permissions, so every guest can connect to it and I
don't need to run modify vswitch grant access every time?
I don't think so. Anyway I don't think you want a vswitch that is access
permission unaware. More
What does lsqeth tell you?
What do you have in /etc/sysconfig/hardware?
Mine is hwcfg-hsi-bus-ccw-0.0
CCW_CHAN_IDS='0.0.0c03 0.0.0c04 0.0.0c05'
CCW_CHAN_MODE=''
CCW_CHAN_NUM='3'
LCS_LANCMD_TIMEOUT=''
MODULE='qeth'
MODULE_OPTIONS=''
QETH_IPA_TAKEOVER='0'
QETH_LAYER2_SUPPORT='0'
QETH_OPTIONS=''
We are running SUSE SLES 10 SP 2. When we login to the server via SSH, our pam
module that validates the userid against Active Directory completes with a
successful logon (as seen on the syslog), but it is some 40-50 seconds before
we see the logon prompt to the user. Can anyone suggest a
I guess I was too shocked by
the idea of letting others tamper with the directory that I was
looking more into recovery from things much worse than this...
And right you are. I wouldn't even trust myself manipulating the directory by
hand. Too much scope for silly mistakes ;-)
Best regards,
The 6500s have a feature called Virtual Switching System (VSS) 1440 that
allows multiple switches to appear as one. (Btw, IEEE 802.3ad requires
awareness of what's going on on all the links. This isn't a limitation of
the VSWITCH.)
We operate in that way and found a major drawback: The physical
We're new to the Linux world. Currently we have 3390 Mod-3 (2.8G) and Mod-9s
(8.4 G) defined to our Linux guests under Z/VM.
On ECKD, same as you mod-3 and mod-9. On FCP terabyte sized luns.
What I'd like to know from the group is - What are your DASD sizes?
Are the complaints about
Cat /proc/scsi/IBMtape?
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390 Port [mailto:linux-...@vm.marist.edu] Namens Lee, Gary D.
Verzonden: woensdag 11 februari 2009 17:07
Aan:
Well, the battle is over, and guess what, FBA won. SCSI/SAS/SATA is
basically all FBA. Wonder what would have happened if IBM hat shut up
Poughkeepsie. Would the world have moved off mainframes in the way it has?
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133
Depending upon what distro you are using, your package manager may allow
you to have both the old and the new version of the package installed at
the same time. If the package is a set of libraries with suffixes, then
that might be sufficient. With RPM-based distros, you tell RPM to install
the
Upgrading a system to current level. Have a vendor product that insists on
the installation of a backlevel component application that causes the
configuration and service management system to report errors in the
configuration. Vendor insists that the backlevel component is the only way,
but the
If you have no other means of discovery, there is a CMS based tool called
SCSDISC. On my system it is located on the CP sample disk and needs some
renaming before use. See the part named SCSIDISC SAMPHELP for instructions. You
attach the FCP device to your CMS user and run it. If there is
This presentation, given at the SHARE Orlando meeting last February, might be
of interest:
Networking with Linux(r) on System z, session 9267, given by Steffen Thoss
(th...@de.ibm.com). He discusses network bonding of multiple NICs in some
detail, and
gives example configurations for SLES 10.
Also
the current level of the bonding driver precludes real balancing of outgoing
data over the nics.
This part I don't agree with. One trap I almost fell into was just looking at
the raw numbers of bytes transmitted for each interface. Then I realized
that a fair amount of traffic flows
The guest-facing side of the VSWITCH does not support LACP.
(Let's avoid terms like port side and trunk side. When a guest is
using a virtual trunk, which side is the trunk side?) LACP is the thing
that lets the host and switch coordinate traffic.
Ok, that's a better terminology, guest-facing
For comparison have a look at cp q nic details and note the difference between
Adapter MAC and Unicast MAC
Anyway, I have set up a production test for tonight that will transfer on the
order of 2-3 TB with about 100 clients. We'll see what will happen there both
balance and performance
A mystery indeed. As Pieter suggested, seeing the output of QUERY NIC
DETAILS would be edifying.
Here is the answer on my system:
TSMSERV:~ # vmcp q nic details
Adapter 0C00.P00 Type: QDIO Name: 0 Devices: 3
MAC: 02-00-00-02-03-96 VSWITCH: SYSTEM QDIOL
RX Packets:
Hello list,
I have been reading in 'Networking with Linux on System z' from Share in last
August (see http://linuxvm.org/Present/SHARE111/S9267kw.pdf) and I see
something that is confusing me. In the part about qeth bonding on page 33 it
says 'Add MAC address to eth0 eth1 (not necessary for
Hello list,
I have been reading in 'Networking with Linux on System z' from Share in
last August (see http://linuxvm.org/Present/SHARE111/S9267kw.pdf) and I see
something that is confusing me. In the part about qeth bonding on page 33 it
says 'Add MAC address to eth0 eth1 (not necessary for
why would you want to channel bond two vnics going to the SAME vswitch?
Chances of a software vnic failure are pretty remote, and may affect all
vnics anyway.
I can see bonding two vnics coupled to two vswitches.
To me it is not a failover issue, but a capacity issue. Eg. with two vnics you
On 1/22/2009 at 4:21 PM, Marcy Cortes marcy.d.cor...@wellsfargo.com
wrote:
We asked about that to IBM via PMR. The answer is no - you cannot do
nic bonding with the vswitch NIC. So I think the presentation is a
little misleading on that page.
Hmm. That's weird. I just set up one on a
None - we asked before we tried. Got a no and so didn't do it.
That's OK though - LACP vswitch was a better way to go and we've since
implemented that.
Marcy,
You just lost me. LACP is on the trunk side of the VSWITCH. I am talking
bonding on the port side of it. What exactly did you think of
I guess I should add that I did it with a Layer 2 VSWITCH, not Layer 3. Since
the bonding module wants to modify the MAC address of the two slave NICs, you
get nastygrams if you try it with Layer 3.
Mark,
From the bonding HOWTO guide:
7. Which switches/systems does it work with?
Mark,
What mode did you set it up with? And do you actually see incoming data on
both eth0 and eth1? If it is only outgoing data you are bonding it will not
help me as Rob pointed out. My problem is with incoming. The 'Linux Ethernet
Bonding Driver HOWTO' on sourceforge suggests you need
Oh, interesting, thanks for the heads-up. We run SAP on SLES10 on z, just
not HA. But it is *stable*, unlike the Windows setup we came from.
Best regards,
Pieter Harder
pieter.har...@brabantwater.nl
tel +31-73-6837133 / +31-6-47272537
-Oorspronkelijk bericht-
Van: Linux on 390
but TSM tape
handling on Linux is so awful and idiosyncratic (not to mention the idiotic
fact that TSM for Linux cannot use channel-attached tape) that the price of
the extra CPU is probably worth it.
I respectfully disagree. I found FCP-based LTO libraries better behaved than
our old 3494.
I respectfully disagree. I found FCP-based LTO libraries better behaved than
our old 3494. And before you ask, I do come from WDSF (aka ADSM V1) and ADSM
V2 and originally bought the 3494 with 3590 for the purpose. The price of the
extra CPU (and licenses) on z/OS will probably outweigh what
1 - 100 of 114 matches
Mail list logo