Moving an IP address between logical interfaces

[Victor Sudakov]

Colleagues,

In the following setup:

$ ifconfig -au
bnx0: flags=201000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4,CoS> mtu 1500 index 2
inet 10.14.134.6 netmask fffffe00 broadcast 10.14.135.255
bnx0:1: flags=201000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4,CoS> mtu 1500 index 2
inet 10.14.140.6 netmask fffffe00 broadcast 10.14.141.255
$

I need to delete 10.14.134.6 making 10.14.140.6 the primary address.

In FreeBSD, all IP aliases are pretty much equal and belong to the
same interface, so any of them can be deleted from the interface
without deleting the others. In Solaris, the aliases belong to
different logical interfaces, so I would have to "move" 10.14.140.6
from bnx0:1 to bnx0, how do I do it?

Thank you in advance for any help.


--
Victor Sudakov, VAS4-RIPE, VAS47-RIPN
2:5005/49@fidonet http://vas.tomsk.ru/

[ji...@specsol.spam.sux.com]

Edit /etc/hosts appropriately.

mv /etc/hostname.bnx0:1 /etc/hostname.bnx0

reboot.

You can do it without rebooting with a series of ifconfig commands, but
you still need to fuss with the files so it sticks across a reboot.

[YTC#1]

With S10 you just need to restart the physical:default service, the age
of reboot is dead.

[ji...@specsol.spam.sux.com]

1. Nowhere was the Solaris version mentioned.

2. Changing the host primary address can have side effects on other running
network services; reboot avoids any.

[Victor Sudakov]

ji...@specsol.spam.sux.com wrote:
> >
> > In the following setup:
> >
> > $ ifconfig -au
> > bnx0: flags=201000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4,CoS> mtu 1500 index 2
> > inet 10.14.134.6 netmask fffffe00 broadcast 10.14.135.255
> > bnx0:1: flags=201000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4,CoS> mtu 1500 index 2
> > inet 10.14.140.6 netmask fffffe00 broadcast 10.14.141.255
> > $
> >
> > I need to delete 10.14.134.6 making 10.14.140.6 the primary address.
> >
> > In FreeBSD, all IP aliases are pretty much equal and belong to the
> > same interface, so any of them can be deleted from the interface
> > without deleting the others. In Solaris, the aliases belong to
> > different logical interfaces, so I would have to "move" 10.14.140.6
> > from bnx0:1 to bnx0, how do I do it?
> >
> > Thank you in advance for any help.

> Edit /etc/hosts appropriately.
>
> mv /etc/hostname.bnx0:1 /etc/hostname.bnx0

> reboot.

What reboot? Are we in comp.os.windows? :) Of course my question
implied without disruption of service.

> You can do it without rebooting with a series of ifconfig commands,

Could you please elaborate on that? I would be most interested in this
series of ifconfig commands. See what I would do in FreeBSD

ifconfig bge0 inet 10.14.134.6 netmask 255.255.254.0
ifconfig bge0 inet 10.14.140.6 netmask 255.255.254.0 alias
(disconnect from 10.14.134.6, connect to 10.14.140.6)
ifconfig bge0 inet 10.14.134.6 delete

Of course those daemons listening on 10.14.134.6 would have to be
restarted but those listening on "*" would survive and work on without
interruption.

> but you still need to fuss with the files so it sticks across a reboot.

Forget about the files, please show how to do the task on the run.

[Victor Sudakov]

Could you please elaborate on that? Some svcadm command I presume?
Yes, it's Solaris 10 5/09.
Let's assume that I have already entered the new 10.14.140.x addresses
to /etc/hostname.bnx0 and /etc/defaultrouter, what's next to eliminate
the old 10.14.134.x ones?

[Victor Sudakov]

ji...@specsol.spam.sux.com wrote:
[dd]

> > With S10 you just need to restart the physical:default service, the age
> > of reboot is dead.

> 1. Nowhere was the Solaris version mentioned.

Sorry. Solaris 10 5/09

> 2. Changing the host primary address can have side effects on other running
> network services; reboot avoids any.

I am willing to restart the services listening on specific addresses.

[YTC#1]

I know, which is why I said "with S10" and not Solaris in general.

>
> 2. Changing the host primary address can have side effects on other running
> network services; reboot avoids any.

Restart those services then, I never reboot unless 100% needed. That is
so antiquated.



--
Bruce Porter
"The internet is a huge and diverse community but mainly friendly"
http://blog.maui.co.uk/index.php/ytc/
There *is* an alternative! http://www.openoffice.org/

[YTC#1]

svcadm restart physical:default

> Let's assume that I have already entered the new 10.14.140.x addresses
> to /etc/hostname.bnx0 and /etc/defaultrouter, what's next to eliminate
> the old 10.14.134.x ones?
>

When you run the restart, it downs the NIC and replumbs it with the new
address. Routing is also handled.

[ji...@specsol.spam.sux.com]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:
> ji...@specsol.spam.sux.com wrote:
> [dd]
>
>> > With S10 you just need to restart the physical:default service, the age
>> > of reboot is dead.
>
>> 1. Nowhere was the Solaris version mentioned.
>
> Sorry. Solaris 10 5/09
>
>> 2. Changing the host primary address can have side effects on other running
>> network services; reboot avoids any.
>
> I am willing to restart the services listening on specific addresses.

Have at it; both ways work.

Personally I find it faster to just reboot, but it is your system.

[ji...@specsol.spam.sux.com]

svcadm restart /network/service

If you do a "svcs -a | grep online | grep network" you will see all the
network services.

Depending on your configuration, others may also need to be restarted.

Which ones?

Like I said, it depends on your configuration, maybe none, maybe several.

That's why I said it is simpler to reboot.

I really don't understand the reluctance to do a reboot.

[Victor Sudakov]

I see.

> > Let's assume that I have already entered the new 10.14.140.x addresses
> > to /etc/hostname.bnx0 and /etc/defaultrouter, what's next to eliminate
> > the old 10.14.134.x ones?
> >

> When you run the restart, it downs the NIC and replumbs it with the new
> address. Routing is also handled.

Will it also unplumb bnx0:1 for me? If it does not, it would be a
pretty kettle of fish with two interfaces configured with 10.14.140.6.

[Cydrome Leader]

or just run /etc/netstart after altering your rc.conf file

[YTC#1]

Depends on what the server is used for, the outage time is *much* lower
for a svcadm restart than a reboot.

I really don't understand your reluctance to not reboot :-)

[ji...@specsol.spam.sux.com]

YTC#1 <b...@ytc1-spambin.co.uk> wrote:
> On 05/20/12 04:57 PM, ji...@specsol.spam.sux.com wrote:

>> svcadm restart /network/service
>>
>> If you do a "svcs -a | grep online | grep network" you will see all the
>> network services.
>>
>> Depending on your configuration, others may also need to be restarted.
>>
>> Which ones?
>>
>> Like I said, it depends on your configuration, maybe none, maybe several.
>>
>> That's why I said it is simpler to reboot.
>>
>> I really don't understand the reluctance to do a reboot.
>>
>
> Depends on what the server is used for, the outage time is *much* lower
> for a svcadm restart than a reboot.

That depends on how much time you have to spend checking other services
to make sure they are OK and restarting them if they are not.

My machines all reboot very quickly.

> I really don't understand your reluctance to not reboot :-)

Of the 129 active services controlled by svcadm, there are only a few where
I would do a reboot.

Changing basic system configuration, such as address and hostname is one
of those few.

[ji...@specsol.spam.sux.com]

Cydrome Leader <pres...@mungepanix.com> wrote:

<snip>

> or just run /etc/netstart after altering your rc.conf file

Neither of which exist on Solaris.

[ji...@specsol.spam.sux.com]

You HAVE to change the files is you want the changes to stay on the next
reboot, whenever it is.

You HAVE to change the files if you want svcadm to restart things the
way you want.

Using svcadm restart is easier than a string of ifconfig commands, but
I didn't mention it in the response as at that time it was not known
if you were running a version of Solaris that supports it.

Also, if you've changed the system name you will have to look at any
configuration file that reference the host name or address, such as
/etc/auto_home.

[Richard B. Gilbert]

Some applications do not allow reboots! Sometimes its not possible! In
other cases its just unreasonably expensive. Twenty years ago, five or
six reboots in an eight hour workday might have been acceptable; you
just said @#$%^&* and pressed "Reset". Now we have PCs in mission
critical applications. We even expect 24x7 from our personal machines.

The only times when my PC is shut down are those when the power fails
for longer than the run time of my UPS.

These days we expect more and we usually get it!

[Ian Collins]

On 05/21/12 06:01 AM, ji...@specsol.spam.sux.com wrote:
> YTC#1<b...@ytc1-spambin.co.uk> wrote:
>> On 05/20/12 04:57 PM, ji...@specsol.spam.sux.com wrote:
>
>>> svcadm restart /network/service
>>>
>>> If you do a "svcs -a | grep online | grep network" you will see all the
>>> network services.
>>>
>>> Depending on your configuration, others may also need to be restarted.
>>>
>>> Which ones?
>>>
>>> Like I said, it depends on your configuration, maybe none, maybe several.
>>>
>>> That's why I said it is simpler to reboot.
>>>
>>> I really don't understand the reluctance to do a reboot.
>>>
>>
>> Depends on what the server is used for, the outage time is *much* lower
>> for a svcadm restart than a reboot.
>
> That depends on how much time you have to spend checking other services
> to make sure they are OK and restarting them if they are not.
>
> My machines all reboot very quickly.

Some x86 servers with SAC controllers and multiple NICs can take a long
time to run through all the BIOS checks. One good reason to use reboot
-f with OpenSolaris derived distributions!

>> I really don't understand your reluctance to not reboot :-)
>
> Of the 129 active services controlled by svcadm, there are only a few where
> I would do a reboot.
>
> Changing basic system configuration, such as address and hostname is one
> of those few.

Maybe on Solaris 10, but definitely not on 11.

--
Ian Collins

[ji...@specsol.spam.sux.com]

Richard B. Gilbert <rgilb...@comcast.net> wrote:

> Some applications do not allow reboots! Sometimes its not possible! In
> other cases its just unreasonably expensive. Twenty years ago, five or
> six reboots in an eight hour workday might have been acceptable; you
> just said @#$%^&* and pressed "Reset". Now we have PCs in mission
> critical applications. We even expect 24x7 from our personal machines.

Oh, nonsense.

There are very, very few applications where a system HAS to be up 24/7
and where those exist, if anyone has any sense at all, there is a backup
fail over system for WHEN, not if, the main goes down.

Many Solaris systems can run for years without a reboot, but eventually
it WILL happen.

For likely 99% of all systems running Solaris, a reboot is a trivial
thing, especially if done at off hours.

And if your system is doing anything at all "important", there is also
24/7 support to do the reboot at off hours.

[Ian Collins]

This is one are where the much maligned nwam stuff in Solaris 11 does
make things easier. I recently had to changes half a dozen addresses on
a one machine to a new subnet. In Solaris 11 this can be done by
creating a new profile and simply activating it. No more editing files.

--
Ian Collins

[ji...@specsol.spam.sux.com]

Ian Collins <ian-...@hotmail.com> wrote:
> On 05/21/12 06:01 AM, ji...@specsol.spam.sux.com wrote:
>> YTC#1<b...@ytc1-spambin.co.uk> wrote:
>>> On 05/20/12 04:57 PM, ji...@specsol.spam.sux.com wrote:
>>
>>>> svcadm restart /network/service
>>>>
>>>> If you do a "svcs -a | grep online | grep network" you will see all the
>>>> network services.
>>>>
>>>> Depending on your configuration, others may also need to be restarted.
>>>>
>>>> Which ones?
>>>>
>>>> Like I said, it depends on your configuration, maybe none, maybe several.
>>>>
>>>> That's why I said it is simpler to reboot.
>>>>
>>>> I really don't understand the reluctance to do a reboot.
>>>>
>>>
>>> Depends on what the server is used for, the outage time is *much* lower
>>> for a svcadm restart than a reboot.
>>
>> That depends on how much time you have to spend checking other services
>> to make sure they are OK and restarting them if they are not.
>>
>> My machines all reboot very quickly.
>
> Some x86 servers with SAC controllers and multiple NICs can take a long
> time to run through all the BIOS checks. One good reason to use reboot
> -f with OpenSolaris derived distributions!

I am unaware of any Solaris systems where there is a BIOS check if you simply
do an OS reboot.

>>> I really don't understand your reluctance to not reboot :-)
>>
>> Of the 129 active services controlled by svcadm, there are only a few where
>> I would do a reboot.
>>
>> Changing basic system configuration, such as address and hostname is one
>> of those few.
>
> Maybe on Solaris 10, but definitely not on 11.

But we are talking about Solaris 10.

Want to through in your thougts on TOPS-20?

[Ian Collins]

On 05/21/12 09:06 AM, ji...@specsol.spam.sux.com wrote:
> Ian Collins<ian-...@hotmail.com> wrote:
>> On 05/21/12 06:01 AM, ji...@specsol.spam.sux.com wrote:
>>> YTC#1<b...@ytc1-spambin.co.uk> wrote:
>>>> On 05/20/12 04:57 PM, ji...@specsol.spam.sux.com wrote:
>>>
>>>>> svcadm restart /network/service
>>>>>
>>>>> If you do a "svcs -a | grep online | grep network" you will see all the
>>>>> network services.
>>>>>
>>>>> Depending on your configuration, others may also need to be restarted.
>>>>>
>>>>> Which ones?
>>>>>
>>>>> Like I said, it depends on your configuration, maybe none, maybe several.
>>>>>
>>>>> That's why I said it is simpler to reboot.
>>>>>
>>>>> I really don't understand the reluctance to do a reboot.
>>>>>
>>>>
>>>> Depends on what the server is used for, the outage time is *much* lower
>>>> for a svcadm restart than a reboot.
>>>
>>> That depends on how much time you have to spend checking other services
>>> to make sure they are OK and restarting them if they are not.
>>>
>>> My machines all reboot very quickly.
>>
>> Some x86 servers with SAC controllers and multiple NICs can take a long
>> time to run through all the BIOS checks. One good reason to use reboot
>> -f with OpenSolaris derived distributions!
>
> I am unaware of any Solaris systems where there is a BIOS check if you simply
> do an OS reboot.

Don't all x86 systems go through BIOS and grub on reboot? Avoiding that
time wastage was the reason for adding the -f option to reboot.

See

http://hub.opensolaris.org/bin/view/Community+Group+on/2008092201

It makes a big difference after an update when booting into the new boot
environment.

--
Ian Collins

[ji...@specsol.spam.sux.com]

While I have not seen every machine that Oracle nee Sun made, none of
them go through the PC BIOS that I have seen using shutdown -i6, which
is what I usually use, not reboot.

They all do the same thing a SPARC machine does.

Keep in mind that I am talking about genuine branded hardware.

The only time I've seen any of these machines go through the BIOS is
when the service processor is rebooted, and that only happens is if you
explicitly reboot it or remove all system power.

Since none-branded hardware doesn't have the service processor, likely
something entirely different happens.

[Ian Collins]

On 05/21/12 10:42 AM, ji...@specsol.spam.sux.com wrote:
> Ian Collins<ian-...@hotmail.com> wrote:
>> On 05/21/12 09:06 AM, ji...@specsol.spam.sux.com wrote:
>>> Ian Collins<ian-...@hotmail.com> wrote:
>
>>>> Some x86 servers with SAC controllers and multiple NICs can take a long
>>>> time to run through all the BIOS checks. One good reason to use reboot
>>>> -f with OpenSolaris derived distributions!
>>>
>>> I am unaware of any Solaris systems where there is a BIOS check if you simply
>>> do an OS reboot.
>>
>> Don't all x86 systems go through BIOS and grub on reboot? Avoiding that
>> time wastage was the reason for adding the -f option to reboot.
>>
>> See
>>
>> http://hub.opensolaris.org/bin/view/Community+Group+on/2008092201
>>
>> It makes a big difference after an update when booting into the new boot
>> environment.
>
> While I have not seen every machine that Oracle nee Sun made, none of
> them go through the PC BIOS that I have seen using shutdown -i6, which
> is what I usually use, not reboot.

Maybe not the PC BIOS (I honestly can't remember), but they (at least
the x4270s I've been using lately) do run through the HBA and NIC BOIS
and grub. Fastboot also works with int 6, from the init man page:

6

Stop the operating system and reboot to the state
defined by the initdefault entry in /etc/inittab.

The service svc:/system/boot-config:default is enabled
by default. When the config/fastreboot_default property
is set to true, init 6 will bypass certain firmware ini-
tialization and test steps, depending on the specific
capabilities of the system.

--
Ian Collins

[ji...@specsol.spam.sux.com]

Ian Collins <ian-...@hotmail.com> wrote:

> Maybe not the PC BIOS (I honestly can't remember), but they (at least
> the x4270s I've been using lately) do run through the HBA and NIC BOIS
> and grub. Fastboot also works with int 6, from the init man page:
>
> 6
>
> Stop the operating system and reboot to the state
> defined by the initdefault entry in /etc/inittab.
>
> The service svc:/system/boot-config:default is enabled
> by default. When the config/fastreboot_default property
> is set to true, init 6 will bypass certain firmware ini-
> tialization and test steps, depending on the specific
> capabilities of the system.

That option isn't in any of the Solaris machines I have access to.

Which version of Solaris is this?

[Ian Collins]

11 and OpenIndiana.

The whole update/reboot process is much improved in newer releases.

--
Ian Collins

[Cydrome Leader]

Wow, really? I was replying to a post where somebody listed commands
they'd run in FreeBSD.

[ji...@specsol.spam.sux.com]

Try reading the original post again, this time for comprehension instead
of kneejerking on isolated words.

[Casper H.S. Dik]

Ian Collins <ian-...@hotmail.com> writes:

>Don't all x86 systems go through BIOS and grub on reboot? Avoiding that
>time wastage was the reason for adding the -f option to reboot.

It's the default these days.

>http://hub.opensolaris.org/bin/view/Community+Group+on/2008092201

>It makes a big difference after an update when booting into the new boot
>environment.

Depends on the hardware; some hardware take little or no time to do the
BIOS checks (e.g., an Intel atom board does it in 2 or 3 seconds)

Casper

[Ian Collins]

You should try the Dell R720 I have been banging my head against recently!

--
Ian Collins

[Casper H.S. Dik]

Ian Collins <ian-...@hotmail.com> writes:

>> Depends on the hardware; some hardware take little or no time to do the
>> BIOS checks (e.g., an Intel atom board does it in 2 or 3 seconds)

>You should try the Dell R720 I have been banging my head against recently!

I never understood why this wasn't part of the design criteria.
(For a MiniITX board, you want to get it booted within seconds, as
it might be used in some automotive equipment)

Casper

[Ian Collins]

I guess you have to leave a couple of second window to allow the user to
enter whatever key is required to ender the BIOS. Then every card with
its own BOIS has to do the same. Naturally they all use a different
key, so the window has to be long enough for the user to read the splash
screen! The the SAS controller has to probe its drives (worse still for
those pesky RAID ones) and remind you what drives are in your box.

I don't know what it does exactly, but the Dell takes ages initialising
its memory, then once all the cards have had their say it runs some form
of system audit. I can usually make a cuppa between a warm start and
the grub menu!

--
Ian Collins

[Victor Sudakov]

Cydrome Leader wrote:

[dd]

> > Could you please elaborate on that? I would be most interested in this
> > series of ifconfig commands. See what I would do in FreeBSD
> >
> > ifconfig bge0 inet 10.14.134.6 netmask 255.255.254.0
> > ifconfig bge0 inet 10.14.140.6 netmask 255.255.254.0 alias
> > (disconnect from 10.14.134.6, connect to 10.14.140.6)
> > ifconfig bge0 inet 10.14.134.6 delete

> or just run /etc/netstart after altering your rc.conf file

This will most likely interrupt IP connectivity. My recipe does not.
Besides, /etc/netstart will try to "start" those network related
scripts in /etc/rc.d while you actually need to "restart" them.

[Casper H.S. Dik]

Ian Collins <ian-...@hotmail.com> writes:

>I guess you have to leave a couple of second window to allow the user to
>enter whatever key is required to ender the BIOS. Then every card with
>its own BOIS has to do the same. Naturally they all use a different
>key, so the window has to be long enough for the user to read the splash
>screen! The the SAS controller has to probe its drives (worse still for
>those pesky RAID ones) and remind you what drives are in your box.

Ah, that's why such well integrated MiniITX board doesn't need to
run other bioses.

>I don't know what it does exactly, but the Dell takes ages initialising
>its memory, then once all the cards have had their say it runs some form
>of system audit. I can usually make a cuppa between a warm start and
>the grub menu!

So, not rebooting then. And they make them for Windows too? Strange.

Casper

[Cydrome Leader]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:
> Cydrome Leader wrote:
>
> [dd]
>
>> > Could you please elaborate on that? I would be most interested in this
>> > series of ifconfig commands. See what I would do in FreeBSD
>> >
>> > ifconfig bge0 inet 10.14.134.6 netmask 255.255.254.0
>> > ifconfig bge0 inet 10.14.140.6 netmask 255.255.254.0 alias
>> > (disconnect from 10.14.134.6, connect to 10.14.140.6)
>> > ifconfig bge0 inet 10.14.134.6 delete
>
>> or just run /etc/netstart after altering your rc.conf file
>
> This will most likely interrupt IP connectivity. My recipe does not.
> Besides, /etc/netstart will try to "start" those network related
> scripts in /etc/rc.d while you actually need to "restart" them.

what are you talking about???? solaris has no /etc/rc.d or /etc/netstart
!!!

while not for everthing, the netstart seems to work ok for some of my
machines. they're all version 4 and 6 though and just web and mailservers.

[YTC#1]

On 05/20/12 09:50 PM, ji...@specsol.spam.sux.com wrote:
> Richard B. Gilbert<rgilb...@comcast.net> wrote:
>
>> Some applications do not allow reboots! Sometimes its not possible! In
>> other cases its just unreasonably expensive. Twenty years ago, five or
>> six reboots in an eight hour workday might have been acceptable; you
>> just said @#$%^&* and pressed "Reset". Now we have PCs in mission
>> critical applications. We even expect 24x7 from our personal machines.
>
> Oh, nonsense.
>
> There are very, very few applications where a system HAS to be up 24/7
> and where those exist, if anyone has any sense at all, there is a backup
> fail over system for WHEN, not if, the main goes down.

You mean a cluster. And unless someone pays out for a fault tolerant
machine there will be an outage.

If there is an outage, someone someone, somewhere will complain.

So, rule 1 = reduce outage time. Don't reboot unless 100% necessary. And
reconfiguring IP does not always need a reboot.

>
> Many Solaris systems can run for years without a reboot, but eventually
> it WILL happen.

Aye, when the system is finally shutdown.

>
> For likely 99% of all systems running Solaris, a reboot is a trivial
> thing, especially if done at off hours.

In some cases, there is no such thing as OOH, hence cluster is used. But
outages still need negotiation.

>
> And if your system is doing anything at all "important", there is also
> 24/7 support to do the reboot at off hours.

As above, OOH, does not always exist.

[ji...@specsol.spam.sux.com]

YTC#1 <b...@ytc1-spambin.co.uk> wrote:
> On 05/20/12 09:50 PM, ji...@specsol.spam.sux.com wrote:
>> Richard B. Gilbert<rgilb...@comcast.net> wrote:
>>
>>> Some applications do not allow reboots! Sometimes its not possible! In
>>> other cases its just unreasonably expensive. Twenty years ago, five or
>>> six reboots in an eight hour workday might have been acceptable; you
>>> just said @#$%^&* and pressed "Reset". Now we have PCs in mission
>>> critical applications. We even expect 24x7 from our personal machines.
>>
>> Oh, nonsense.
>>
>> There are very, very few applications where a system HAS to be up 24/7
>> and where those exist, if anyone has any sense at all, there is a backup
>> fail over system for WHEN, not if, the main goes down.
>
> You mean a cluster. And unless someone pays out for a fault tolerant
> machine there will be an outage.

Well, a cluster is one way to do it, but not the only way.

It depends on what the system is being used for and what it is doing.

> If there is an outage, someone someone, somewhere will complain.

Yeah, OK, if it is so important, there had better be back up.

> So, rule 1 = reduce outage time. Don't reboot unless 100% necessary. And
> reconfiguring IP does not always need a reboot.

That part is at least true, if the system is that important.

>> Many Solaris systems can run for years without a reboot, but eventually
>> it WILL happen.
>
> Aye, when the system is finally shutdown.

When something breaks, and things ALWAYS break, even on genuine branded
systems.

If the system is that important and there isn't back up for when, not if,
something breaks, those running the system are fools.

>> For likely 99% of all systems running Solaris, a reboot is a trivial
>> thing, especially if done at off hours.
>
> In some cases, there is no such thing as OOH, hence cluster is used. But
> outages still need negotiation.

Again, clusters are not the only way to ensure up time.

And "negotiation" is often little more than an email that the system
will reboot at 3AM unless someone has a problem with that.

>> And if your system is doing anything at all "important", there is also
>> 24/7 support to do the reboot at off hours.
>
> As above, OOH, does not always exist.

Out of Hours??

So what; it does for most systems.

If you are running a system so blazingly important that it can not be down
for a couple of minutes any time during the week or any time during the
day and you don't have a backup system of some kind, good luck to you.

BTW, even if you change the IP address manually, it will still take a couple
of minutes from the time you start until the time the machine is running
on the new address.

During that time your system is totally not network accessable.

[Victor Sudakov]

Cydrome Leader wrote:
> >
> > [dd]
> >
> >> > Could you please elaborate on that? I would be most interested in this
> >> > series of ifconfig commands. See what I would do in FreeBSD
> >> >
> >> > ifconfig bge0 inet 10.14.134.6 netmask 255.255.254.0
> >> > ifconfig bge0 inet 10.14.140.6 netmask 255.255.254.0 alias
> >> > (disconnect from 10.14.134.6, connect to 10.14.140.6)
> >> > ifconfig bge0 inet 10.14.134.6 delete
> >
> >> or just run /etc/netstart after altering your rc.conf file
> >
> > This will most likely interrupt IP connectivity. My recipe does not.
> > Besides, /etc/netstart will try to "start" those network related
> > scripts in /etc/rc.d while you actually need to "restart" them.

> what are you talking about???? solaris has no /etc/rc.d or /etc/netstart !!!

We were certainly talking about FreeBSD, see above.

> while not for everthing, the netstart seems to work ok for some of my
> machines. they're all version 4 and 6 though and just web and mailservers.

[Victor Sudakov]

YTC#1 wrote:

> >
> > Many Solaris systems can run for years without a reboot, but eventually
> > it WILL happen.

> Aye, when the system is finally shutdown.

[sudakov@admin ~] /usr/bin/ruptime -t | grep sun2
sun2 up 1557+03:56, 0 users, load 0.46, 0.48, 0.51

[ji...@specsol.spam.sux.com]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:
> YTC#1 wrote:
>
>> >
>> > Many Solaris systems can run for years without a reboot, but eventually
>> > it WILL happen.
>
>> Aye, when the system is finally shutdown.
>
> [sudakov@admin ~] /usr/bin/ruptime -t | grep sun2
> sun2 up 1557+03:56, 0 users, load 0.46, 0.48, 0.51
>

Most people concider running rwhod a security risk plus the following from
the man page:

WARNINGS
This service can cause network performance problems when
used by several hosts on the network. It is not run at most
sites by default.

[ji...@specsol.spam.sux.com]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:

> We were certainly talking about FreeBSD, see above.

Who is "we" and you do realize this is a Solaris group?

[Cydrome Leader]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:
> Cydrome Leader wrote:
>> >
>> > [dd]
>> >
>> >> > Could you please elaborate on that? I would be most interested in this
>> >> > series of ifconfig commands. See what I would do in FreeBSD
>> >> >
>> >> > ifconfig bge0 inet 10.14.134.6 netmask 255.255.254.0
>> >> > ifconfig bge0 inet 10.14.140.6 netmask 255.255.254.0 alias
>> >> > (disconnect from 10.14.134.6, connect to 10.14.140.6)
>> >> > ifconfig bge0 inet 10.14.134.6 delete
>> >
>> >> or just run /etc/netstart after altering your rc.conf file
>> >
>> > This will most likely interrupt IP connectivity. My recipe does not.
>> > Besides, /etc/netstart will try to "start" those network related
>> > scripts in /etc/rc.d while you actually need to "restart" them.
>
>> what are you talking about???? solaris has no /etc/rc.d or /etc/netstart !!!
>
> We were certainly talking about FreeBSD, see above.

I was joking. some idiot here doesn't understand that.

[Victor Sudakov]

ji...@specsol.spam.sux.com wrote:
> >> >
> >> > Many Solaris systems can run for years without a reboot, but eventually
> >> > it WILL happen.
> >
> >> Aye, when the system is finally shutdown.
> >
> > [sudakov@admin ~] /usr/bin/ruptime -t | grep sun2
> > sun2 up 1557+03:56, 0 users, load 0.46, 0.48, 0.51
> >
>
> Most people concider running rwhod a security risk

Not in a trusted intranet, really.

> plus the following from
> the man page:

> WARNINGS
> This service can cause network performance problems when
> used by several hosts on the network. It is not run at most
> sites by default.

This man page must have been written aeons ago, before the
introduction of FastEthernet. Besides, the man page says also

"If used, include the -m multicast option."

and that's exactly what I do.

[Victor Sudakov]

ji...@specsol.spam.sux.com wrote:

> > We were certainly talking about FreeBSD, see above.

> Who is "we"

"Cydrome Leader" and myself, I guess.

> and you do realize this is a Solaris group?

Comparing Solaris with other operating systems is not prohibited, even
if Solaris loses the competition :)

[ji...@specsol.spam.sux.com]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:
> ji...@specsol.spam.sux.com wrote:
>> >> >
>> >> > Many Solaris systems can run for years without a reboot, but eventually
>> >> > it WILL happen.
>> >
>> >> Aye, when the system is finally shutdown.
>> >
>> > [sudakov@admin ~] /usr/bin/ruptime -t | grep sun2
>> > sun2 up 1557+03:56, 0 users, load 0.46, 0.48, 0.51
>> >
>>
>> Most people concider running rwhod a security risk
>
> Not in a trusted intranet, really.

There really is no such thing.

>> plus the following from
>> the man page:
>
>> WARNINGS
>> This service can cause network performance problems when
>> used by several hosts on the network. It is not run at most
>> sites by default.
>
> This man page must have been written aeons ago, before the
> introduction of FastEthernet. Besides, the man page says also
>
> "If used, include the -m multicast option."
>
> and that's exactly what I do.

Well, good for you, you read the instructions.

I have one other question, if there are 0 users and no load on the
machine, how can it be so important it has to be up 24/7?

[ji...@specsol.spam.sux.com]

Victor Sudakov <v...@mpeks.no-spam-here.tomsk.su> wrote:
> ji...@specsol.spam.sux.com wrote:
>
>> > We were certainly talking about FreeBSD, see above.
>
>> Who is "we"
>
> "Cydrome Leader" and myself, I guess.
>
>> and you do realize this is a Solaris group?
>
> Comparing Solaris with other operating systems is not prohibited, even
> if Solaris loses the competition :)

Non sequitur.

[YTC#1]

On 05/26/12 06:41 PM, ji...@specsol.spam.sux.com wrote:
> Victor Sudakov<v...@mpeks.no-spam-here.tomsk.su> wrote:
>> ji...@specsol.spam.sux.com wrote:
>>>>>>
>>>>>> Many Solaris systems can run for years without a reboot, but eventually
>>>>>> it WILL happen.
>>>>
>>>>> Aye, when the system is finally shutdown.
>>>>
>>>> [sudakov@admin ~] /usr/bin/ruptime -t | grep sun2
>>>> sun2 up 1557+03:56, 0 users, load 0.46, 0.48, 0.51
>>>>
>>>
>>> Most people concider running rwhod a security risk
>>
>> Not in a trusted intranet, really.
>
> There really is no such thing.

Aye, damn those pesky internal hackers ! :-)

Or are you referring to the fact that the intranet does not exist (in
which case I agree with you)

>
>>> plus the following from
>>> the man page:
>>
>>> WARNINGS
>>> This service can cause network performance problems when
>>> used by several hosts on the network. It is not run at most
>>> sites by default.
>>
>> This man page must have been written aeons ago, before the
>> introduction of FastEthernet. Besides, the man page says also
>>
>> "If used, include the -m multicast option."
>>
>> and that's exactly what I do.
>
> Well, good for you, you read the instructions.
>
> I have one other question, if there are 0 users and no load on the
> machine, how can it be so important it has to be up 24/7?

Because he wants to use use it at 0 seconds usage ? :-)

[ji...@specsol.spam.sux.com]

YTC#1 <b...@ytc1-spambin.co.uk> wrote:
> On 05/26/12 06:41 PM, ji...@specsol.spam.sux.com wrote:
>> Victor Sudakov<v...@mpeks.no-spam-here.tomsk.su> wrote:
>>> ji...@specsol.spam.sux.com wrote:
>>>>>>>
>>>>>>> Many Solaris systems can run for years without a reboot, but eventually
>>>>>>> it WILL happen.
>>>>>
>>>>>> Aye, when the system is finally shutdown.
>>>>>
>>>>> [sudakov@admin ~] /usr/bin/ruptime -t | grep sun2
>>>>> sun2 up 1557+03:56, 0 users, load 0.46, 0.48, 0.51
>>>>>
>>>>
>>>> Most people concider running rwhod a security risk
>>>
>>> Not in a trusted intranet, really.
>>
>> There really is no such thing.
>
> Aye, damn those pesky internal hackers ! :-)

Not too likely for small operations, I agree, but a lot more prevelant
in large operations than most people would like to admit.

Especially in very large operations with multiple locations.

I have seen large, very carefully controlled, internal networks brought
down because some doofus brought in a USB stick, stuck it into their
Windows workstation to show their co-worker something, not knowing the
stick was also infected with something that spread to the other Windows
machines and then started attacking the servers on the internal net.

This kind of thing doesn't usually get publicized outside the organization.

> Or are you referring to the fact that the intranet does not exist (in
> which case I agree with you)

Actually, a bit of both.

>>>> plus the following from
>>>> the man page:
>>>
>>>> WARNINGS
>>>> This service can cause network performance problems when
>>>> used by several hosts on the network. It is not run at most
>>>> sites by default.
>>>
>>> This man page must have been written aeons ago, before the
>>> introduction of FastEthernet. Besides, the man page says also
>>>
>>> "If used, include the -m multicast option."
>>>
>>> and that's exactly what I do.
>>
>> Well, good for you, you read the instructions.
>>
>> I have one other question, if there are 0 users and no load on the
>> machine, how can it be so important it has to be up 24/7?
>
> Because he wants to use use it at 0 seconds usage ? :-)

Har!!

[Cydrome Leader]

I was talking about his recently. What if anything would happen to a super
old internet facing unix box? Same for a windows 95 machine.

are there script kiddies looking for sunos machines to compromise?

Is somebody really going to find root at the machine and discover big deal
information?

If I had an IPX laying around, I'd try it.

[Victor Sudakov]

ji...@specsol.spam.sux.com wrote:
> >>
> >> Most people concider running rwhod a security risk
> >
> > Not in a trusted intranet, really.

> There really is no such thing.

There is no absolute trust (perhaps only in God), but there are
different security levels, and there is risk assessment. For certain
scenarios we can consider a net "trusted".

> >> plus the following from
> >> the man page:
> >
> >> WARNINGS
> >> This service can cause network performance problems when
> >> used by several hosts on the network. It is not run at most
> >> sites by default.
> >
> > This man page must have been written aeons ago, before the
> > introduction of FastEthernet. Besides, the man page says also
> >
> > "If used, include the -m multicast option."
> >
> > and that's exactly what I do.

> Well, good for you, you read the instructions.

> I have one other question, if there are 0 users and no load on the
> machine, how can it be so important it has to be up 24/7?

I was not talking about this particular system. This one with the 1561
days uptime is a web-server for some SCADA system, I could reboot it
any time I wanted. It's just fun to have an uptime like that.