Re: [smartos-discuss] zone-specific (?) commands not autocompleting,

[Nahum Shalman]

As an anecdote, perhaps not super-helpful, I just upgraded my home machine
to joyent_20170413T062226Z earlier today and am not seeing that issue. I
don't recall seeing it on joyent_20170330T015208Z and in theory the version
I was running before that was that same joyent_20170315T210557Z where I
also do not recall seeing that issue.

That bash prompt looks different from what I would expect to see based on
the /root/.bashrc that SmartOS ships (
https://github.com/joyent/smartos-live/blob/master/overlay/generic/root/.bashrc
not showing any new changes since 2015...)

-Nahum

On Thu, Apr 20, 2017 at 1:37 PM, Matthew Parsons <
matt...@aspergerexperts.com> wrote:

> joyent_20170315T210557Z
>
> vmadm sta [tab]
> vmadm stop (first 4 UUID) [tab],
> zlogin (first 4 or so UUID) [tab], etc does not work.
>
> (Regular tab-completion for file paths and such work)
>
> Additionally, "vmadm" by itself outputs:
>
> -bash-4.3# vmadm
> op=
> uuid=
> Requested unique lookup but found 0 results.
> Alias  not found.
>
> Usage: /usr/sbin/vmadm  [options]
>
> (Regular usage info follows)
>
> Using 'which' command, verified that zlogin is /usr/sbin/zlogin, vmadm
> is /usr/sbin/vmadm -> ../vm/sbin/vmadm
>
> I don't know where to go from here, other than a reboot...
>
>
> *smartos-discuss* | Archives
> 
>  |
> Modify
> 
> Your Subscription 
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Windows KVM on ZFS Questions

[Nahum Shalman]

I do not work for Joyent, but I've run a lot of SmartOS and I have opinions
that I think are relevant (inline):

· Does joyent set “sync=disabled” on windows zvol’s?
>

I highly doubt it. Joyent hardware specs that I have seen are either
all-SSD, or have SSD SLOG devices in the pool to ensure snappy ZFS
synchronous writes.
Changing that setting makes things "go faster" at the very real cost of
opening up the potential for the guest VM to experience data corruption of
the NTFS filesystem in the case of power-failure of the SmartOS host system.
That is a tradeoff that you can make if you know for sure that you have
other protections in place (For example if they are all thin-provisioned
from a generic image, and user data are all either unimportant, backed up,
or preserved with regular snapshotting, etc.).
Joyent can't know that about the workloads in the JPC, so they depend on
fast SLOGs, not playing fast and loose with customer data.


> · Does joyent use “dedup=on” on cloud nodes?
>

I highly doubt this one as well. dedup has some serious drawbacks outside
of some very specific use cases. They leverage thin provisioning more so
than dedup. I do not recommend use of dedup without lots of research into
the consequences and serious testing.


> · Is there a list of “best practices” options to set on windows
> VM’s to get the most performance? Taking into account that the ZVOL will
> have ntfs on top of it?
>

I'm not sure. Certainly getting the virtio drivers working correctly for
the disk helps.


> · Is there a performance hit using thin-provisioned ZVOL’s?
>

I'm pretty sure it's the opposite and that the ARC is able to detect that
everyone is using the same block and only needs to cache one.


> · Is LZ4 compression okay on a ZVOL dataset? Is there a
> performance impact to the VM’s or the GZ?
>

My experience has generally been that LZ4 is an all-around win. Your CPUs
can do the LZ4 compression so fast and the reduction of how much needs to
actually land on disk, combined with the improved disk utilization make it
something I generally enable by default.


> · Are there any performance options that ive missed here such as
> record size that could impact performance positively (or negatively)?
>

By default the zvol's get the 8k block size. At this time I don't think
that SmartOS advertises an 8k block size to the guest. It's possible that
there could be some interesting performance analysis that one could to
tweaking both the zvol block size and the block size advertised by QEMU to
the guest, but I have not done so.

I hope that someone from Joyent will weigh in on anything that I've either
gotten wrong, or where their recommendations are different from mine.

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Note to help installs

[Nahum Shalman]

Looks like a small step in the right direction landed and is in the current
release:
https://github.com/joyent/smartos-live/commit/457eeeb#diff-02843e11b99b4466766211737b6c2cb6R787

(Thanks Robert!)

-Nahum

On Fri, Mar 17, 2017 at 6:12 PM, Steve <smar...@szmidt.org> wrote:

> On 03/17/2017 17:30, Nahum Shalman wrote:
> > Which brings me back to my original question. Do you think if instead of:
> >
> > Please manually create/import a zpool named zones.
> >
> > It said:
> >
> > Please manually create/import a zpool named 'zones'.
> >
> > That the quotation marks would have drawn your attention a bit more?
> >
> > The fact that you missed it might be a sign of a small UX problem that
> > is both real and addressable.
> 
> Sure. If it was me I probably would have said something like:
> 
> VITAL! Ensure a zpool zone called 'zones' is created/imported.
> 
> or
> 
> IMPORTANT! A zpool named 'zones' must be created/imported.
> 
> I like putting the priority first to highlight the importance. Makes it
> harder to miss. Had I fumbled onto a line like that I'd probably get it
> in spite of the cloud of confusion I was in at the time.
> 
> --
> Steve
> 



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Docker endpoint in SmartOS

[Nahum Shalman]

Just for clarity:
1. Yes you need a full set of Triton APIs to run the Docker API endpoint.
It speaks the Docker API on its front and implements its backend with the
Triton APIs [1]
2. Yes, the easiest way to set up a test environment for the Docker API is
probably to stand up CoaL.

If the reason you're asking about this is for work on getting SaltStack to
play nicely with Triton, I encourage you to reply in the affirmative and I
encourage someone at Joyent to simply set you up with an account in the
public cloud for testing purposes.

-Nahum

P.S. If you have an existing SmartOS host I have an old blog post on how to
run CoaL in a KVM on SmartOS here:
https://blog.shalman.org/running-sdc-coal-on-smartos/

[1] Code is here:
https://github.com/joyent/sdc-docker/tree/master/lib/backends/sdc There's
an abandoned "lx-zone" backend in the codebase but it's incomplete and no
one is working on it.

On Sat, Mar 25, 2017 at 12:00 PM, C. R. Oldham  wrote:

> Sorry to be dense, and installing CoAL is the easiest way to do that?
>
> --cro
>
>
> > On Mar 25, 2017, at 08:11, Coy Hile  wrote:
> >
> > I believe you need Triton stack for that. It is Triton that provides the
> API for docker.
> >
> > Sent from my iPhone
> >
> >> On Mar 24, 2017, at 21:56, C. R. Oldham  wrote:
> >>
> >> Greetings,
> >>
> >> Is there a way to provide the Docker endpoint inside a SmartOS
> installation without installing or running all of CoAL?  I'm having trouble
> figuring out if there are a few packages I can install to get the bare
> minimum of sdc-* commands, or if CoAL is really the minimum set of what I
> would need.
> >>
> >> Thanks,
> >>
> >> --cro
> >>
> >> 
> >>
> >
> >
> 
> 



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Note to help installs

[Nahum Shalman]

Which brings me back to my original question. Do you think if instead of:

Please manually create/import a zpool named zones.

It said:

Please manually create/import a zpool named 'zones'.

That the quotation marks would have drawn your attention a bit more?

The fact that you missed it might be a sign of a small UX problem that is
both real and addressable.

-Nahum


On Fri, Mar 17, 2017 at 2:40 PM, Steve <smar...@szmidt.org> wrote:

> On 03/17/2017 14:03, Nahum Shalman wrote:
> > The link I sent was to the source code of the installer.
> > When you run the installer and choose manual pool layout, it
> > specifically tells you to create/import a pool name zones.
> 
> Well maybe I cannot read! :)
> 
> My attention was all on how do I manually configure ZFS and simply did
> not duplicate the 'a pool named zones'.
> 
> OK, right where I would expect it to be it says 'a zpool named zones.' I
> clearly only read the 'create/import' part which I knew and skipped the
> rest of the line. (Jikes!)
> 
> Thanks for entertaining my ignorance! :) Hopefully I'll read it slower
> next time... :\
> 
> --
> Steve
> 



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] vioblk driver

[Nahum Shalman]

On Mon, Feb 13, 2017 at 4:15 PM, Robert Mustacchi <r...@joyent.com> wrote:

> On 2/13/17 12:56 , Youzhong Yang wrote:
> > It seems it's not that simple. I will leave it for later troubleshooting
> as
> > I don't have cycles to debug it now.
>
> Do you have a write up somewhere about what you encountered and what
> went wrong?
>

I think you should be able to reproduce the issues encountered by booting
SmartOS under SmartOS KVM with qemu_extra_opts tweaking the virtio physical
and logical block sizes e.g.:

On Mon, Jan 30, 2017 at 2:07 PM, Nahum Shalman <nshal...@omniti.com> wrote:

> I think you should be able to set "qemu_extra_opts" to e.g. "-global
> virtio-blk-pci.physical_block_size=8192 -global 
> virtio-blk-pci.logical_block_size=4096"
> to test tweaking those values.
>

I was certainly able to trip a bug with those particular settings which
seemed like the kind of thing an SSD might do.
When using those settings, diskinfo got the size wrong and the system was
unable to create a ZFS pool.

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] vioblk driver

[Nahum Shalman]

Only setting:
"qemu_extra_opts": "-global virtio-blk-pci.physical_block_size=4096"
appeared to work. diskinfo reported the correct size, creating a pool on it
succeeded, and it had an ashift of 12.

Not sure where that will leave you.
-Nahum


On Mon, Jan 30, 2017 at 2:30 PM, Nahum Shalman <nshal...@omniti.com> wrote:

> I wonder if I've just stumbled on a bug. I provisioned a VM with a zvol of
> 8GB, and with those qemu_extra_opts, but "diskinfo" thinks the disk is 64GB
> in size...
> I don't know the full mdb-fu you used to find that address and dump out
> those properties.
>
> Something seems a little fishy here.
> -Nahum
>
> On Mon, Jan 30, 2017 at 2:07 PM, Nahum Shalman <nshal...@omniti.com>
> wrote:
>
>> That's only modifying the volblocksize of the underlying zvol. Getting
>> QEMU to advertise a different blocksize might be trickier which is probably
>> what's necessary.
>> I found reference to "-global ide-drive.physical_block_size=4096" in
>> http://wiki.qemu.org/download/qemu-doc.html#sec_005finvocation
>>
>> After a bunch of digging I ran the following two commands and got the
>> output as shown:
>>
>> # /smartdc/bin/qemu-system-x86_64 -device ? 2>&1 | grep -i virtio
>> name "virtio-balloon-pci", bus PCI
>> name "virtio-serial-pci", bus PCI, alias "virtio-serial"
>> name "virtio-net-pci", bus PCI
>> name "virtio-blk-pci", bus PCI, alias "virtio-blk"
>> name "virtconsole", bus virtio-serial-bus
>> name "virtserialport", bus virtio-serial-bus
>> # /smartdc/bin/qemu-system-x86_64 -device virtio-blk-pci,? 2>&1 | grep
>> block_size
>> virtio-blk-pci.logical_block_size=uint16
>> virtio-blk-pci.physical_block_size=uint16
>>
>> So I think you should be able to set "qemu_extra_opts" to e.g. "-global
>> virtio-blk-pci.physical_block_size=8192 -global 
>> virtio-blk-pci.logical_block_size=4096"
>> to test tweaking those values.
>>
>> Good luck!
>> -Nahum
>>
>>
>> On Mon, Jan 30, 2017 at 12:11 PM, Youzhong Yang <youzh...@gmail.com>
>> wrote:
>>
>>> Thanks, it works.
>>>
>>> However, I am having difficulty setting the block size to anything other
>>> than 512:
>>>
>>> vmadm tells the block size is 8192,
>>>
>>> # vmadm get b175850d-92fd-cf81-d352-90d40e5076ce |json disks
>>> [
>>>   {
>>> "path": "/dev/zvol/rdsk/zones/b175850d-92fd-cf81-d352-90d40e5076ce-d
>>> isk1",
>>> "boot": false,
>>> "model": "virtio",
>>> "media": "disk",
>>> "zfs_filesystem": "zones/b175850d-92fd-cf81-d352
>>> -90d40e5076ce-disk1",
>>> "zpool": "zones",
>>> "size": 65536,
>>> "compression": "on",
>>> "refreservation": 65536,
>>> "block_size": 8192
>>>   }
>>> ]
>>>
>>> but inside the VM, it says the block size is 512:
>>>
>>> > ff03d8df9700::print -t struct vioblk_softc sc_virtio.sc_features
>>> sc_blk_size sc_pblk_size
>>> uint32_t sc_virtio.sc_features = 0x1454
>>> int sc_blk_size = 0x200
>>> int sc_pblk_size = 0x200
>>>
>>> Quite confused ..
>>>
>>> Any thoughts?
>>>
>>> Thanks,
>>>
>>> --Youzhong
>>>
>>>
>>>
>>> On Mon, Jan 30, 2017 at 10:10 AM, Nahum Shalman <nshal...@omniti.com>
>>> wrote:
>>>
>>>> You can also create a SmartOS KVM VM on a SmartOS machine with a virtio
>>>> virtual disk.
>>>>
>>>> -Nahum
>>>>
>>>> On Mon, Jan 30, 2017 at 9:38 AM, Youzhong Yang <youzh...@gmail.com>
>>>> wrote:
>>>>
>>>>> Hi All,
>>>>>
>>>>> A quick question - how can I make use of vioblk device driver?
>>>>> Creating a SmartOS VM using Linux KVM?
>>>>>
>>>>> The reason I am asking this is that I want to test a change in blkdev,
>>>>> to exercise some code path in it which can't be done using NVMe devices,
>>>>> such as the 'dump' functionality.
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Youzhong
>>>>>
>>>>>
>>>>
>>> *smartos-discuss* | Archives
>>> <https://www.listbox.com/member/archive/184463/=now>
>>> <https://www.listbox.com/member/archive/rss/184463/28443469-fb954443> |
>>> Modify
>>> <https://www.listbox.com/member/?;>
>>> Your Subscription <http://www.listbox.com>
>>>
>>
>>
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] vioblk driver

[Nahum Shalman]

I wonder if I've just stumbled on a bug. I provisioned a VM with a zvol of
8GB, and with those qemu_extra_opts, but "diskinfo" thinks the disk is 64GB
in size...
I don't know the full mdb-fu you used to find that address and dump out
those properties.

Something seems a little fishy here.
-Nahum

On Mon, Jan 30, 2017 at 2:07 PM, Nahum Shalman <nshal...@omniti.com> wrote:

> That's only modifying the volblocksize of the underlying zvol. Getting
> QEMU to advertise a different blocksize might be trickier which is probably
> what's necessary.
> I found reference to "-global ide-drive.physical_block_size=4096" in
> http://wiki.qemu.org/download/qemu-doc.html#sec_005finvocation
>
> After a bunch of digging I ran the following two commands and got the
> output as shown:
>
> # /smartdc/bin/qemu-system-x86_64 -device ? 2>&1 | grep -i virtio
> name "virtio-balloon-pci", bus PCI
> name "virtio-serial-pci", bus PCI, alias "virtio-serial"
> name "virtio-net-pci", bus PCI
> name "virtio-blk-pci", bus PCI, alias "virtio-blk"
> name "virtconsole", bus virtio-serial-bus
> name "virtserialport", bus virtio-serial-bus
> # /smartdc/bin/qemu-system-x86_64 -device virtio-blk-pci,? 2>&1 | grep
> block_size
> virtio-blk-pci.logical_block_size=uint16
> virtio-blk-pci.physical_block_size=uint16
>
> So I think you should be able to set "qemu_extra_opts" to e.g. "-global
> virtio-blk-pci.physical_block_size=8192 -global 
> virtio-blk-pci.logical_block_size=4096"
> to test tweaking those values.
>
> Good luck!
> -Nahum
>
>
> On Mon, Jan 30, 2017 at 12:11 PM, Youzhong Yang <youzh...@gmail.com>
> wrote:
>
>> Thanks, it works.
>>
>> However, I am having difficulty setting the block size to anything other
>> than 512:
>>
>> vmadm tells the block size is 8192,
>>
>> # vmadm get b175850d-92fd-cf81-d352-90d40e5076ce |json disks
>> [
>>   {
>> "path": "/dev/zvol/rdsk/zones/b175850d-92fd-cf81-d352-90d40e5076ce-
>> disk1",
>> "boot": false,
>> "model": "virtio",
>> "media": "disk",
>> "zfs_filesystem": "zones/b175850d-92fd-cf81-d352-90d40e5076ce-disk1",
>> "zpool": "zones",
>> "size": 65536,
>> "compression": "on",
>> "refreservation": 65536,
>> "block_size": 8192
>>   }
>> ]
>>
>> but inside the VM, it says the block size is 512:
>>
>> > ff03d8df9700::print -t struct vioblk_softc sc_virtio.sc_features
>> sc_blk_size sc_pblk_size
>> uint32_t sc_virtio.sc_features = 0x1454
>> int sc_blk_size = 0x200
>> int sc_pblk_size = 0x200
>>
>> Quite confused ..
>>
>> Any thoughts?
>>
>> Thanks,
>>
>> --Youzhong
>>
>>
>>
>> On Mon, Jan 30, 2017 at 10:10 AM, Nahum Shalman <nshal...@omniti.com>
>> wrote:
>>
>>> You can also create a SmartOS KVM VM on a SmartOS machine with a virtio
>>> virtual disk.
>>>
>>> -Nahum
>>>
>>> On Mon, Jan 30, 2017 at 9:38 AM, Youzhong Yang <youzh...@gmail.com>
>>> wrote:
>>>
>>>> Hi All,
>>>>
>>>> A quick question - how can I make use of vioblk device driver? Creating
>>>> a SmartOS VM using Linux KVM?
>>>>
>>>> The reason I am asking this is that I want to test a change in blkdev,
>>>> to exercise some code path in it which can't be done using NVMe devices,
>>>> such as the 'dump' functionality.
>>>>
>>>> Thanks,
>>>>
>>>> Youzhong
>>>>
>>>>
>>>
>> *smartos-discuss* | Archives
>> <https://www.listbox.com/member/archive/184463/=now>
>> <https://www.listbox.com/member/archive/rss/184463/28443469-fb954443> |
>> Modify
>> <https://www.listbox.com/member/?;>
>> Your Subscription <http://www.listbox.com>
>>
>
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] vioblk driver

[Nahum Shalman]

That's only modifying the volblocksize of the underlying zvol. Getting QEMU
to advertise a different blocksize might be trickier which is probably
what's necessary.
I found reference to "-global ide-drive.physical_block_size=4096" in
http://wiki.qemu.org/download/qemu-doc.html#sec_005finvocation

After a bunch of digging I ran the following two commands and got the
output as shown:

# /smartdc/bin/qemu-system-x86_64 -device ? 2>&1 | grep -i virtio
name "virtio-balloon-pci", bus PCI
name "virtio-serial-pci", bus PCI, alias "virtio-serial"
name "virtio-net-pci", bus PCI
name "virtio-blk-pci", bus PCI, alias "virtio-blk"
name "virtconsole", bus virtio-serial-bus
name "virtserialport", bus virtio-serial-bus
# /smartdc/bin/qemu-system-x86_64 -device virtio-blk-pci,? 2>&1 | grep
block_size
virtio-blk-pci.logical_block_size=uint16
virtio-blk-pci.physical_block_size=uint16

So I think you should be able to set "qemu_extra_opts" to e.g.
"-global virtio-blk-pci.physical_block_size=8192
-global virtio-blk-pci.logical_block_size=4096" to test tweaking those
values.

Good luck!
-Nahum


On Mon, Jan 30, 2017 at 12:11 PM, Youzhong Yang <youzh...@gmail.com> wrote:

> Thanks, it works.
>
> However, I am having difficulty setting the block size to anything other
> than 512:
>
> vmadm tells the block size is 8192,
>
> # vmadm get b175850d-92fd-cf81-d352-90d40e5076ce |json disks
> [
>   {
> "path": "/dev/zvol/rdsk/zones/b175850d-92fd-cf81-d352-
> 90d40e5076ce-disk1",
> "boot": false,
> "model": "virtio",
> "media": "disk",
> "zfs_filesystem": "zones/b175850d-92fd-cf81-d352-90d40e5076ce-disk1",
> "zpool": "zones",
> "size": 65536,
> "compression": "on",
> "refreservation": 65536,
> "block_size": 8192
>   }
> ]
>
> but inside the VM, it says the block size is 512:
>
> > ff03d8df9700::print -t struct vioblk_softc sc_virtio.sc_features
> sc_blk_size sc_pblk_size
> uint32_t sc_virtio.sc_features = 0x1454
> int sc_blk_size = 0x200
> int sc_pblk_size = 0x200
>
> Quite confused ..
>
> Any thoughts?
>
> Thanks,
>
> --Youzhong
>
>
>
> On Mon, Jan 30, 2017 at 10:10 AM, Nahum Shalman <nshal...@omniti.com>
> wrote:
>
>> You can also create a SmartOS KVM VM on a SmartOS machine with a virtio
>> virtual disk.
>>
>> -Nahum
>>
>> On Mon, Jan 30, 2017 at 9:38 AM, Youzhong Yang <youzh...@gmail.com>
>> wrote:
>>
>>> Hi All,
>>>
>>> A quick question - how can I make use of vioblk device driver? Creating
>>> a SmartOS VM using Linux KVM?
>>>
>>> The reason I am asking this is that I want to test a change in blkdev,
>>> to exercise some code path in it which can't be done using NVMe devices,
>>> such as the 'dump' functionality.
>>>
>>> Thanks,
>>>
>>> Youzhong
>>>
>>>
>>
> *smartos-discuss* | Archives
> <https://www.listbox.com/member/archive/184463/=now>
> <https://www.listbox.com/member/archive/rss/184463/28443469-fb954443> |
> Modify
> <https://www.listbox.com/member/?;>
> Your Subscription <http://www.listbox.com>
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] vioblk driver

[Nahum Shalman]

You can also create a SmartOS KVM VM on a SmartOS machine with a virtio
virtual disk.

-Nahum

On Mon, Jan 30, 2017 at 9:38 AM, Youzhong Yang  wrote:

> Hi All,
>
> A quick question - how can I make use of vioblk device driver? Creating a
> SmartOS VM using Linux KVM?
>
> The reason I am asking this is that I want to test a change in blkdev, to
> exercise some code path in it which can't be done using NVMe devices, such
> as the 'dump' functionality.
>
> Thanks,
>
> Youzhong
>
> *smartos-discuss* | Archives
> 
>  |
> Modify
> 
> Your Subscription 
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] 'Defrag' a KVM disk

[Nahum Shalman]

On Wed, Jan 18, 2017 at 3:25 AM, Ian Collins  wrote:

> On 01/18/17 08:31 PM, Matthias Goetzke wrote:
>
> We have a machine which temporarily got 98% full and the boot time of a
> KVM windows server has gotten quite extreme (minutes on the black part of
> the boot alone).
>
> Apart from over-allocating, we guess it might have to do with
> fragmentation of the underlying C drive (just a hunch due to the number of
> writes on C and the fact that overall fragmentation was temporarily >78%)
> since another win machine still booted up quite quickly.
>
>
>
> If a kvm disk were just a file then I could just make a copy of the C
> drive (I have enough space now) into a new file which should (in theory)
> just create new aligned blocks and once I delete the old file it should be
> defragmented mostly.
>
>
> A KVM disk is a sparse provisioned zfs volume, not a file.
>
> Part of the basic maintenance I perform on Windows KVMs is well documented
> here:
>
> https://www.maketecheasier.com/shrink-your-virtualbox-vm.
>
> You will see the REFER number for the volume reduce dramatically when you
> run sdelete.
>
> It is also fairly easy to increase the volsize of the underlying disk and
> expand your windoze partition to fill it.
>

+1 for this method with a particular caveat:
Turn on compression (I recommend LZ4) for your KVM disks, whether by simply
enabling it on the entire zones pool, or on the VM disk directly.

If compression is off, writing out blocks of zeros won't free up space in
the pool. With compression on, blocks of zeros are highly compressible and
thus do free up space.
This of course also requires deleting any snapshots you may have made that
refer to the old garbage on disk.

Step 1: Check if compression is enabled for the disk with "zfs get
compression zones/-disk"
Step 1.5: If not, turn it on with "zfs set compression=lz4
zones/-disk"
Step 2: Inside the VM, write out lots of zeros (I'm not familiar with
sdelete, but I'd imagine it works)
Step 3: Verify that you've successfully freed up disk space.

Incidentally, I have a feeling that if you zeroed things out with
compression off, you might need to write new garbage and then write new
zeros after turning compression on to free up the disk space.

For others reading this thread who use something like Linux, when I've done
this I generally just use "dd" reading from /dev/zero inside the VM to
write out a file full of zeros to get the same effect.
When using ZFS inside the VM as I have sometimes done, make sure to turn
compression off on the filesystem inside the VM so that it's actually
writing out lots of zeros for the zvol compression to notice. "zpool create
-o compression=off /deleteme; dd if=/dev/zero bs=128k
of=//deleteme; zfs destroy /deleteme"

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] LX Zone from CD/ISO Image

[Nahum Shalman]

Peter beat me to the punch, but to put one detail a bit more succinctly:
If the Sophos Firewall UTM depends on things like iptables functionality
from the Linux kernel then it will not work with the LX brand at all.
Sadly, I suspect very strongly that it does depend on iptables.

-Nahum


On Mon, Jan 16, 2017 at 12:24 PM, Peter Kelm 
wrote:

> Mat,
>
> You’d be (more or less) limited to a KVM zone if all you have is an ISO of
> an appliance. In a nutshell an LX zone is a native zone („Solaris“) using a
> Linux compatibility layer. It is not a full Linux OS, does not run its own
> kernel…
>
> Theoretically you could build an LX zone that mimics the setup of that
> Sophos appliance.
> 1) Determine what distribution (and specific version) the Sophos appliance
> is based on. Then start out with an LX dataset of the same distribution (on
> SmartOS).
> 2) „Diff“ the filesystems and replicate all changes (install packages,
> edit config files,…) on that LX zone/machine.
>
> I looked at this a while ago for an AV appliance but found that it is too
> cumbersome and completely unsupported anyway. Fortunately our AV vendor
> also offered an RPM install in addition to the appliance ISO. So I spun up
> an Ubuntu LX zone using the Joyent provided dataset and installed those
> RPMs per the directions or the AV software supplier…
>
> Let me know if this makes it clearer.
>
> Peter
>
> Am 16.01.2017 um 17:57 schrieb Mat Schreiber via smartos-discuss <
> smartos-discuss@lists.smartos.org>:
>
>
>
> Hi,
>
> sorry if my question is answered somewhere in the Joyent Wikis, but I
> found nothing clearly explaining it:
>
> I want to install the Sophos  Firewall UTM in an LX Zone (as I assume LX
> Zones to be ways faster than KVM Zones).
> So far so good. But I do not know how to create a VM now, as I don't have
> an empty Image in imgadm.
> I just have instructions from Oracle wiht LX Zones and installation from
> CD:
> Installing and Booting lx Branded Zones (System Administration Guide:
> Oracle Solaris Containers-Resource Management and Oracle Solaris Zones)
> 
>
> But I am not sure if it would be possible afterwards getting this thing
> again under control of vmadm...
>
> Or is it like that, that I should take an image (which one wouldn't
> matter) and when set up I do an unconfigure to that Zone and install it
> again from CD as described in the Oracle Instruction...
> What to do best?
>
> Sorry and thanks,
> Mat
>
>
>
>
>
> *smartos-discuss* | Archives
> 
>  |
> Modify
> 
> Your Subscription 
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] fs-local is too rigorous

[Nahum Shalman]

On Fri, Jan 6, 2017 at 3:28 AM, Gernot Straßer 
wrote:

> This is regarding ./proto/lib/svc/method/fs-local
>
> I was playing around with zfs crypto when I figured that script will fail
> if there is an encrypted dataset waiting for passphrase input.
>
> In line 90 it does a zfs mount -va and fails hard if there is an error,
> which causes all dependent services to fail . This includes sshd which
> renders  the system unaccessible.
>
> My proposal would be to skip checking for mount errors, but there might be
> reasons I am not aware of.
>

What if you alter the "canmount" property of all your encrypted filesystems
to be "noauto"?
Since they require interactivity to be mounted, that's probably pretty
accurate, and I think should exclude them from "zfs mount -va"

I could imagine making the argument that "canmount=noauto" should be the
default for encrypted filesystems...

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] vnc resolution configurable?

[Nahum Shalman]

Wow, resurrecting a thread from 2013 with my email address from my previous
employer. Okay, let's jump in.

On Thu, Dec 8, 2016 at 3:47 AM, Fred Liu <fred_...@issi.com> wrote:

> We noticed that GUI since SL7.2 did not work with the low resolution
> 1024x768 which is the default-detected resolution from SmartOS vnc server.
>

That might have to do with the default amount of video memory QEMU gives
the VM. You may be able to change that with qemu_extra_opts. I've only ever
done that to the QXL device rather than the default VGA device so I don't
know what the variable you'd need to set would be.

Is there a way to configure the vnc to run in higher resolution like
> 1280x1024? And I don't need to hack Linux to forcedly run in higher
> resolution.
>

I think what you're asking is "How do I make all changes on the
SmartOS/QEMU/Client side so that I don't have to make any changes within
the guest OS?"

The first thing I would try to do is force the VNC client to try to connect
at a higher resolution. That might be enough to jolt the X server up to the
higher resolution so that the login UI will be able to render correctly.

Figuring out what arguments to add to qemu_extra_opts is what I would
explore second.



> BTW, is spice stable enough on SmartOS?
>

My old email address which was CCd on your message was from my time at
Elysium Digital which was acquired by Stroz Friedberg[1] in 2015 and which
I left in March of this year (2016).
When I left, ~50 people in that Boston office were still using Spice
enabled VMs (running Fedora as the guest OS) on SmartOS as their primary
work systems every day which they had been doing for years.
I would have argued then that for some value of stable, that counted.

I do not know if that system is still in use, and even if it is, it
probably won't be for much longer.
For a while after I left I was still regularly using a Spice enabled Fedora
VM on my home server for most of my personal stuff.
At this point most of what I was doing on it has been split apart between a
Mac laptop and a Centos 7 LX branded zone on the SmartOS machine.
Which means that the Spice for SmartOS work is currently dormant in terms
of my time and I'm pretty sure no one else is actively developing it.

So it may be somewhat stable, but it's kind of rotting on the vine at this
point.


If anyone out there is using the Spice enabled QEMU code and needs support,
my current employer, OmniTI, is a consulting company and can sell you my
time to work on it.


-Nahum

[1] Stroz Friedberg has since been acquired by Aon


> -Original Message-
> > From: Nahum Shalman [mailto:nshalman-smar...@elys.com]
> > Sent: 星期一, 三月 25, 2013 21:01
> > To: smartos-discuss@lists.smartos.org
> > Cc: Alain O'Dea
> > Subject: Re: [smartos-discuss] Cannot get working VNC Viewer onto KVM
> Zone
> >
> > On 03/23/2013 09:56 PM, Alain O'Dea wrote:
> > > Is SPICE a feasible option here?  I notice the experimental
> > > configuration options for it.  I'm really only looking at having a
> > > console long enough to install the OS and bootstrap a Bomgar session
> > > onto the desktop.
> >
> > Feasible, possibly. Ideal, probably not.
> > In order to use Spice, you need to use an alternate QEMU binary rather
> than
> > the one that ships with the platform.
> > (See http://wiki.smartos.org/display/DOC/Spice+on+SmartOS for details)
> >
> > The benefit of using "my" QEMU is that you can use Spice. The biggest
> down
> > side is that if you run into problems, you're not using the Joyent
> supported
> > QEMU.
> >
> > One option if you're trying to build a fresh image to use later as a
> template that
> > will have SSH already working, you could try using the Spice enabled
> QEMU for
> > the stage of building your image, verify that a fresh boot of that image
> does
> > bring up SSH as desired, then use that image with the platform QEMU.
> >
> > So, order of preference from best to worst from my perspective:
> >Try to get VNC working.
> >Use Spice enabled QEMU for image building, platform QEMU for
> production
> > workload.
> >Use Spice enabled QEMU for building and running in production
> >
> > I do run "my" QEMU in production, but I can't provide support.
> >
> > -Nahum
> >
> > P.S.  I might not-so-secretly be hoping that you'll choose #3 *and* that
> you'll
> > then ask Joyent to support it... A guy can dream.
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] New host and CPU question

[Nahum Shalman]

I like to think about CPU as a resource either under contention (more
processes need CPU time than CPU cores to go around) or not:

CPU caps limit how much you can use when there's no contention. In the
Joyent Could it ensures people don't get more CPU time than they pay for.
(example 1 below)

CPU shares ensure fairness when there is contention. If CPU shares are
doled out in proportion to memory doled out (Like in the Joyent Cloud),
when everyone is fighting for CPU, the more memory you're paying for, the
more CPU cycles you'll get. (example 2 below)

If all the zones on a box belong to the same customer (as might be the case
on your machine), you could set the cpu caps much higher, and depend on the
shares for fairness.

Examples for clarification:

1. I fire up a g4-highcpu-2G zone in the Joyent cloud. It gets a CPU cap of
100. It happens to land on a freshly deployed compute node with no other
zones on it.
I then run 2 CPU intensive processes. There is no contention so the machine
is totally capable of scheduling each of them on a separate CPU and running
them full throttle.
However, because I'm only paying for 1 1vCPU I hit that cap of 100 and my
two processes have to take turns running. If my workload is really that CPU
intensive I should scale the zone up to g4-highcpu-4G which gets 2vCPUs and
a cap of 200.

2. At home I have a machine with 8 cores running SmartOS. I don't put caps
on my zones because I am the only customer on the box so to speak. I
provision 3 zones on the machine: the "important" zone that gets 600
shares, and two "unimportant" ones that each get 200 shares. Each of them
is running 8 CPU intensive processes (apparently I'm a glutton for
punishment). I have given out 1 shares. Because my numbers all divide
evenly I can use simple fractions. The important zone get 6/10 or 3/5 of
the CPU, and the other two each get 2/10 or 1/5. My math adds up: 3/5 + 1/5
+ 1/5 = 5/5 =1. The important zone gets approximately 4.8 cores worth of
CPU time, and the unimportant ones each get approximately 1.6 cores worth.
This is as good as things can be when I foolishly run too many CPU
intensive workloads on my machine at home. Maybe it's time for a beefier
box. :-P

Was this helpful?
-Nahum

On Fri, Dec 2, 2016 at 3:47 AM, Len Weincier  wrote:

> Just to note I see that in triton the cpu_cap is set as well as cpu_shares
> == max_physical_memory
>
> I am struggling to understand how this works. I am guessing that if the
> ratio of cpu_cap to cores is low then its not that noticeable to the
> guests, the threads get through in a reasonable amount of time but with the
> E7-x cpu's that have high core counts and the ratio is higher this will
> become a more serious problem especially for multithreaded apps ?
>
> Thanks
> Len
>
>
> On 2 December 2016 at 09:44, Len Weincier  wrote:
>
>> This is the classic symptom of a machine with a cpu_cap less than its
>>> CPU (core) count.
>>>
>>
>> So the vm sees all the cores and dispatches threads / processes for that
>> many cores but the actual uasage is limited by the cpu_cap so the threads
>> are stalling waiting for the OS to give them a slot to run hence the load
>> avg climbing iiuc. Setting cpu_cap=0 sets the machine free and its
>> responsive
>>
>> Whats the right way to share the cores for LX brand machines then seeing
>> as we cant limit the access to the cores ? In triton is see the default
>> packages set the cpu_cap as well which would result in the same issue ?
>>
>> I have had a look at https://wiki.smartos.org/di
>> splay/DOC/CPU+Caps+and+Shares  and am not sure what todo ?
>>
>> Thanks
>> Len
>>
>>
> *smartos-discuss* | Archives
> 
>  |
> Modify
> 
> Your Subscription 
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] ln command behavior on LX zones

[Nahum Shalman]

In each of those three environments you should run "alias ll" and you will
probably see that those are different in each environment.
On the SmartOS GZ, that alias "is coming from root's ".bashrc" file
In the other two environments it is coming from something in those.
It could be reasonable to request that the SmartOS GZ alias be made to
match one of those other two (or both if they are the same).

On my SmartOS box I see:
# grep alias /root/.bashrc
alias ll='ls -lF'
alias ls='ls --color=auto'

-Nahum

On Thu, Nov 17, 2016 at 10:18 AM, Giorgio Valoti  wrote:

> Hi all,
> I’ve noticed this anomalous behavior with the ln command on LX zones. I
> don’t know which of the two behavior is compliant.
>
>
> From a test GZ:
> > > ssh coal
> > Warning: Permanently added '10.99.99.7' (ECDSA) to the list of known
> hosts.
> > - SmartOS (build: 20161113T161122Z)
> > [root@headnode (dev) ~]# cd /tmp
> > [root@headnode (dev) /tmp]# mkdir test && ln -s test dest
> > [root@headnode (dev) /tmp]# ll
> > total 16
> > lrwxrwxrwx   1 root root   4 nov 17 13:29 dest -> test/
> > -rw-r--r--   1 root root   0 nov 17 07:17 iconf_entries.2811
> > drwxr-xr-x   2 root root 117 nov 17 13:29 test/
>
>
> Via zlogin (same with, say, docker exec):
> > [root@headnode (dev) /tmp]# zlogin c6a989d2-717d-43d6-bf66-ac8a9b8bb7b6
> > [Connected to zone 'c6a989d2-717d-43d6-bf66-ac8a9b8bb7b6' pts/4]
> > Last login: Wed Nov 16 11:14:21 UTC 2016 from localhost on pts/5
> > root@c6a989d2717d:~# cd /tmp
> > root@c6a989d2717d:/tmp# mkdir test && ln -s test dest
> > root@c6a989d2717d:/tmp# ll
> > total 43991
> > drwxrwxrwt  8 rootroot  28 Nov 17 13:30 ./
> > drwxr-xr-x 26 rootuucp  30 Nov 16 15:17 ../
> > lrwxrwxrwx  1 rootroot   4 Nov 17 13:30 dest -> test/
> > drwxr-xr-x  2 rootroot   2 Nov 17 13:30 test/
> >
>
>
>
> Here’s the same version of Ubuntu (running on VirtualBox)
> > root@dev/tmp > mkdir test && ln -s test dest
> > root@dev/tmp > ll
> > total 52K
> > drwxr-xr-x 25 rootroot4.0K Nov 16 11:02 ..
> > drwxr-xr-x  2 rootroot4.0K Nov 17 14:38 test
> > lrwxrwxrwx  1 rootroot   4 Nov 17 14:38 dest -> test
> > drwxrwxrwt  8 rootroot 12K Nov 17 14:38 .
> 
> 
> --
> Giorgio Valoti
> 



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] About install pkgin in global zone

[Nahum Shalman]

On Wed, Nov 16, 2016 at 1:37 PM, Tiraen  wrote:

> If manually substitute name of the package set:
>
> [root@ds821695 /]# curl -kO https://pkgsrc.joyent.com/
> packages/SmartOS/bootstrap/bootstrap-2016Q3-tools-20161020.tar.gz
>   % Total% Received % Xferd  Average Speed   TimeTime Time
>  Current
>  Dload  Upload   Total   SpentLeft
>  Speed
>  20 61.6M   20 12.3M0 0  4538k  0  0:00:13  0:00:02  0:00:11
> 5105kcurl: (23) Failed writing body (12624 != 16384)
>

You're trying to download it to a location in the root filesystem which is
the ramdisk and doesn't have enough space.
You'll want to download it to somewhere like /opt or /zones

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] ZFS / ZPOOL usage understanding

[Nahum Shalman]

On Wed, Nov 16, 2016 at 3:24 AM, Len Weincier  wrote:

> 2. Is compression only to improve I/O but not file system capacity ?
>

To get a sense of how much compression is helping you in terms of capacity
you can look at the output of:
zfs list -o
name,logicalreferenced,referenced,refcompressratio,logicalused,used,compressratio

The zfs(1m) man page explains what those mean pretty clearly. There's a
copy of it online here: https://smartos.org/man/1m/zfs

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] SmartOS build flag day

[Nahum Shalman]

I've closed off https://github.com/joyent/smartos-live/pull/529 as it
appears to be addressed by https://smartos.org/bugview/OS-5651 in
https://github.com/joyent/smartos-live/commit/13f5ba1

I've accordingly updated
https://wiki.smartos.org/display/DOC/Building+SmartOS+on+SmartOS though I
haven't actually had a chance yet to confirm the fix. I'll try to do that
soon.

I am curious whether / when the preferred build zone image will change to
15.4.x multiarch.

-Nahum

On Thu, Nov 10, 2016 at 12:43 PM, Robert Mustacchi  wrote:

> Hi,
> 
> If you don't build SmartOS, then you can ignore this e-mail.
> 
> We just put back "OS-5652 platform build needs to work with python 2.7".
> This unblocks building on newer base images. However, it requires that
> you have updates to smartos-live, illumos-joyent, and illumos-extra.
> 
> To pass this flag day, you must run the following in each workspace that
> you have:
> 
> $ gmake clobber
> $ gmake update
> $ ./configure
> 
> If you have any questions, please let me know.
> 
> Thanks,
> Robert
> 



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Re: Sdc on smartos

[Nahum Shalman]

On Fri, Oct 7, 2016 at 7:53 AM, Bouktaib Mohssin  wrote:

> i've tried this tutorial too http://blog.shalman.org/
> running-sdc-coal-on-smartos/ but still the same problem with the kvm
>

My blog post, let's see if I can help.


>
> On Fri, Oct 7, 2016 at 10:47 AM, Bouktaib Mohssin <
> bouktaib.mohs...@gmail.com> wrote:
>
>> Hello
>>
>> I'm trying to install sdc on smartos following the instructions of this
>> tutorial https://wiki.smartos.org/display/DOC/Install+SDC+on+SmartOS ,
>> when i start the sdc-headone virtual machine the vm shutdown unexpectedly
>> hence i can log to continue the installation process
>>
>> # zfs snapshot zones/${HNUUID}-disk0@sdc-fixed
>> # vmadm start ${HNUUID}
>> # vmadm console ${HNUUID}
>>
>> [root@ibms3550 /tmp]# vmadm list UUID TYPE RAM STATE ALIAS
>> 06da7ec1-6f08-c278-9eee-d49de42d0669 KVM 8192 stopped sdc-headnode
>>
>
I ran through my blog post on my home machine (the delay in getting back to
you was finding time to free up enough disk space to be able to do the test
again...)
and I actually ran into a similar problem.

When I mounted the disk0 zvol and examined the grub menu.lst file, it
appeared corrupt.
I reran qemu-img convert and mounted it up before taking the snapshot.
The second time it worked.
I don't know how/why qemu-img would have messed up copying the image to the
zvol.

The symptom with the corrupt image was that it sat at the console but
nothing ever rendered.
After I re-did the imaging onto the zvol and verified that the grub config
was valid everything booted fine.

Please try again, but verify that the menu.lst file that get's cleaned with
sed is readable before booting.

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Troubleshooting "Invalid JSON payload: Unexpected string" (LX Zone)

[Nahum Shalman]

Yup, looks like the missing comma was my fault... Should be fixed now:
https://wiki.smartos.org/pages/diffpages.action?pageId=3604602=5373984

As for getting a wiki account, you'll have to get one of the Joyent folks
to set you up with that.

Sorry for the confusion!
-Nahum

On Thu, Sep 22, 2016 at 12:44 PM, Matthew Parsons <
matt...@aspergerexperts.com> wrote:

>
>
> On Thu, Sep 22, 2016 at 1:24 AM, Johan Claesson  wrote:
>
>>
>>> I'm attempting to create a pretty bog-standard LX zone, copying and
>>> barely modifying the example from https://wiki.smartos.org/displ
>>> ay/DOC/LX+Branded+Zones [1] like so:
>>>
>>>

>
>> The problem is you're missing a comma on line 12.
>>
>> Tip: Use the json command to get more help regarding syntax errors.
>>
>> Eg. json -f /path/to/json_file
>> json: error: input is not JSON: Expected ',' instead of '"' at line 13,
>> column 5:
>> "primary": true
>> 
>>
>
> Excellent, that worked, thank you! (My first time using json, so not
> familiar w/ it, closest seems to be yaml which I use(d) w/ Puppet.)
>
> I checked the Wiki entry, the last edit that affected it was by a Mr.
> Shalman - I attempted to edit it, but there doesn't appear to be a location
> to create/request an account.
>
>
>
> *smartos-discuss* | Archives
> 
>  |
> Modify
> 
> Your Subscription 
>



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] the partition size is fixed 1.86G in USB stick. Can it be enlarged?

[Nahum Shalman]

On 02/22/2016 01:55 AM, Paul Sture wrote:

On 19 Feb 2016, at 14:42, Nahum Shalman wrote:


After resizing you'd probably need to re-install grub...
An alternative is to format the USB stick with a ZFS pool and boot 
that way (as a bonus you can then take snapshots and I've heard that 
it boots a little faster than pcfs...):

http://blog.shalman.org/smartos-zfs-boot-media/

If you want to stick with pcfs, you could follow similar steps to 
format your USB stick at its full capacity and then install grub on it.


The reason that the image is fixed at such a small size is so that 
people can use it on pretty much any USB stick they have lying around 
(If we made the image 8GB and a person had lots of 4GB sticks lying 
around etc.)


Is it possible to restrict the size of the image to just slightly less 
than the size of a given USB stick?


I ask because when I was trying out napp-it, that came as a 16GB image 
and out of half a dozen USB sticks I had here, only one was the full 
size (I ended up using a 32GB stick for a backup).  The dirty truth 
about USB sticks is that they are often a few MB short of their 
advertised capacity*.  That doesn't matter much when using them as 
general storage, but it does when you are trying to put a fixed size 
image onto one.


* with the USB sticks I had available at the time, the shortfall 
varied from 9MB to 273MB




That's in fact what we do: 
https://github.com/joyent/smartos-live/blob/master/tools/build_iso#L179
The image is 20 bytes long. I've never heard reports that it was 
still too big, but if anyone has had trouble with it they should speak up.


Also, if any vendor is shipping something smaller than that and labeling 
it as 2GB they should be publicly shamed. That is the absolute smallest 
number that can acceptably be called 2GB.


-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] enable network/physical in noinstall/recovery?

[Nahum Shalman]

On 02/22/2016 04:05 AM, Fred Liu wrote:

From: Nahum Shalman [mailto:nshalman-smar...@elys.com]
Sent: 星期五, 二月 19, 2016 21:55
To: smartos-discuss@lists.smartos.org
Subject: Re: [smartos-discuss] enable network/physical in noinstall/recovery?

If you know the mac address of the NIC you want to bring online, you can edit 
the kernel command line within grub and append the following stuff:

,admin_mac=11:22:33:44:55:66,admin_ip=dhcp
or
,admin_mac=11:22:33:44:55:66,admin_ip=192.168.x.y,admin_netmask=255.255.255.0,admin_gateway=192.168.x.1

[Fred]: I tried. It seems it did not work.


That's what I get for doing things from memory. it's "admin_nic" not 
admin_mac... Sorry about that.




If you can't edit the kernel command line, the alternative to getting the system to finish coming 
up is to manually create the /usbkey directory, and create a minimal /usbkey/config file with the 
networking configuration you want. After that "sysinfo -f" then "svcadm clear 
network/physical" and the system should finish bringing services online.

You'll probably need to fix up /etc/resolv.conf if you want to make outbound 
connections.

[Fred]: It works well!


Good.


Be aware that you're bringing the machine online with a known root password 
that will work over SSH. You can replace the hash in the kernel command line 
with something else if you want to be extra cautious.

I hope that helps.

[Fred]: Are there comprehensive documents about all the parameters fed to 
kernel? Could the alternative way be
 Adding ",smartos=false,noimport=true" to the normal boot kernel?

To answer your original question about why there isn't a rescue mode that 
includes networking, I think the main reasons are:
1. SmartOS is often used on machines with multiple NICs. There's no way to know 
which to use without more information from the operator.
2. SmartOS is often used on networks with no DHCP server. Without being told 
what IP address to use there would be no way for SmartOS to configure working 
networking.
3. The system would then come online with a known root password and SSH running 
and that could be dangerous.
4. In light of those other reasons, no one has written code for it.

[Fred]: Hm, what is your way to restore a bad zpool like zones?


I've rarely had issues that I couldn't clear up with a few commands 
directly at the console. Most of my machines either have IPMI (which I 
can use for serial-over-lan) or are plugged into an IP KVM so I've 
generally not needed to have networking come up in rescue mode.


When I do need a machine to come up with networking in rescue mode I use 
one of the two above methods. I usually use the manual creation of 
/usbkey/config because I don't have the MAC address handy until after 
I've logged in. I suppose it could be nice to write a script that would 
take a named interface like "igb0" that could automate bringing 
networking up, but it's so rare that I need rescue mode that I haven't 
felt the need to automate it.


-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] enable network/physical in noinstall/recovery?

[Nahum Shalman]

If you know the mac address of the NIC you want to bring online, you can 
edit the kernel command line within grub and append the following stuff:


,admin_mac=11:22:33:44:55:66,admin_ip=dhcp
or
,admin_mac=11:22:33:44:55:66,admin_ip=192.168.x.y,admin_netmask=255.255.255.0,admin_gateway=192.168.x.1

If you can't edit the kernel command line, the alternative to getting 
the system to finish coming up is to manually create the /usbkey 
directory, and create a minimal /usbkey/config file with the networking 
configuration you want. After that "sysinfo -f" then "svcadm clear 
network/physical" and the system should finish bringing services online.


You'll probably need to fix up /etc/resolv.conf if you want to make 
outbound connections.


Be aware that you're bringing the machine online with a known root 
password that will work over SSH. You can replace the hash in the kernel 
command line with something else if you want to be extra cautious.


I hope that helps.

To answer your original question about why there isn't a rescue mode 
that includes networking, I think the main reasons are:
1. SmartOS is often used on machines with multiple NICs. There's no way 
to know which to use without more information from the operator.
2. SmartOS is often used on networks with no DHCP server. Without being 
told what IP address to use there would be no way for SmartOS to 
configure working networking.
3. The system would then come online with a known root password and SSH 
running and that could be dangerous.

4. In light of those other reasons, no one has written code for it.

-Nahum

On 02/19/2016 08:03 AM, Fred Liu wrote:


Yeah. That is doable. But because there is no zpool imported, 
network/physical can’t be started. And a lot of protocols can’t run 
neither.


Thanks.

Fred

*From:*Jorge Schrauwen [mailto:sjorge...@blackdot.be]
*Sent:* 星 期五, 二月19, 2016 19:21
*To:* smartos-discuss@lists.smartos.org
*Cc:* Fred Liu
*Subject:* Re: [smartos-discuss] enable network/physical in 
noinstall/recovery?


You can use ifconfig to bring it up manually.

ifconfig e1000g0 inet auto-dhcp (from memory so it may be incorrect)

On 2016-02-19 11:30, Fred Liu wrote:

Hi,

“noinstall/recovery” is really helpful to fix/save zpools. But it
is offline. Is it hard to enable networking in it? Just like
“fail-safe-mode-with-network”.

Nobody wants to recover from remote backup but that is the ideal life.

Thanks.

Fred

*smartos-discuss* | Archives 
 
 
| Modify 
 
Your Subscription 	[Powered by Listbox] 







---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] the partition size is fixed 1.86G in USB stick. Can it be enlarged?

[Nahum Shalman]

After resizing you'd probably need to re-install grub...
An alternative is to format the USB stick with a ZFS pool and boot that 
way (as a bonus you can then take snapshots and I've heard that it boots 
a little faster than pcfs...):

http://blog.shalman.org/smartos-zfs-boot-media/

If you want to stick with pcfs, you could follow similar steps to format 
your USB stick at its full capacity and then install grub on it.


The reason that the image is fixed at such a small size is so that 
people can use it on pretty much any USB stick they have lying around 
(If we made the image 8GB and a person had lots of 4GB sticks lying 
around etc.)


-Nahum


On 02/18/2016 08:10 PM, Fred Liu wrote:

I have got no success with gparted.


Thanks.
Fred

Sent from Outlook Mobile 

_
From: Jorge Schrauwen >

Sent: 星期二, 二月 16, 2016 8:08 下午
Subject: RE: [smartos-discuss] the partition size is fixed 1.86G in 
USB stick. Can it be enlarged?
To: >

Cc: Fred Liu >


You can also just resize the fat partition after writing to the UBS stick.

Jorge

On 2016-02-16 12:58, Fred Liu wrote:

It should be doable if we can hack in “gmake usb”.

*From:*Micky [mailto:mickylmar...@gmail.com]
*Sent:* 星期二, 二月16, 2016 19:06
*To:* smartos-discuss@lists.smartos.org

*Subject:* Re: [smartos-discuss] the partition size is fixed 1.86G
in USB stick. Can it be enlarged?

Likely no, cause it's a disk image.

On Tue, Feb 16, 2016 at 1:20 PM, Fred Liu > wrote:



You can also just resize the fat partition after writing to the UBS stick.

Jorge

On 2016-02-16 12:58, Fred Liu wrote:


It should be doable if we can hack in “gmake usb”.

*From:*Micky [mailto:mickylmar...@gmail.com]
*Sent:* 星期二, 二月16, 2016 19:06
*To:* smartos-discuss@lists.smartos.org
*Subject:* Re: [smartos-discuss] the partition size is fixed 1.86G in 
USB stick. Can it be enlarged?


Likely no, cause it's a disk image.

On Tue, Feb 16, 2016 at 1:20 PM, Fred Liu > wrote:


*smartos-discuss* | Archives 
 
 
| Modify 
 
Your Subscription 	[Powered by Listbox] 







---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Question regarding package sizing

[Nahum Shalman]

Trent Mick pointed me to these two links on IRC:
https://smartos.org/bugview/PAPI-127
https://github.com/joyent/sdc-vmapi/blob/master/lib/common/validation.js#L948-L964

Looks like "fss" gets turned into "cpu_shares" and it also looks like if 
you don't provide it when you define the package it gets calculated 
based on memory sizing...


Still no idea on the rest.

-Nahum

On 02/16/2016 12:36 PM, Nahum Shalman wrote:
At this point you may want to CC the sdc-discuss mailing list; I don't 
know the innards of SDC like I know SmartOS...


I have a vague recollection that in general Joyent (and thus SDC by 
default?) assign cpu_shares relative to max_physical_memory.
I can't find anything definitive from a little bit of spelunking in 
the code though, so don't quote me on that one...


I did notice that PAPI has no notion of cpu_shares, but VMAPI and 
vm-agent do. Somewhere a translation happens but I haven't found it 
yet and need to stop looking for now.


Again, maybe someone on the sdc-discuss mailing list would know where 
to look...


I'd love to know what the answer is though, so please follow up on 
this thread if/when you find out.


-Nahum

On 02/16/2016 07:43 AM, Benjamin Bergia wrote:

Thank you, it does make more sense now.
It also looks like I have been doing it all wrong since the beginning.

Three points are still bothering me and I would like to get a 
confirmation if anybody as any idea.


First, in the papi documentation 
<https://github.com/joyent/sdc-papi/blob/master/docs/index.md#formulas>, 
there is some obscure calculations regarding the cpu cap. Are these 
formulas up to date ? If yes what is the "burst ratio" used to 
calculate the cpu_burst_ratio.


Then, the same documentation talk about cpu_burst_ratio, fss, 
ram_ratio. After creating a new package, these values are not 
returned by the papi. This let me think that the documentation might 
not be up to date. But I would still be interested by getting a 
confirmation.


And finally, the triton documentation 
<https://docs.joyent.com/private-cloud/packages/configuring#packagedefinitions> specify 
that vCPU is only used by KVM and not required for SmartOS zones.


I run exclusively SmartOS zones, that's why I am wondering if in my 
case, there is really some calculation running behind the scene to 
generate a cpu_share value from a given amount of vCPUs. Or is the 
vCPU field wrongly marked as "required" in the AdminUI and SDC's 
packages contain 1 vCPU as a simple place holder ?



On Mon, Feb 15, 2016 at 10:51 PM, Jorge Schrauwen 
<sjorge...@blackdot.be <mailto:sjorge...@blackdot.be>> wrote:


    Excellent explenation Nahum!




On 2016-02-15 15:39, Nahum Shalman wrote:

On 02/15/2016 01:30 AM, Benjamin Bergia wrote:

Hi,

I recently noticed that all the packages used by SDC
zones are using some strange settings. All of them, even
database ones, are using 1 vCPU with a cap of 400. I can
picture in my head the "meaning" of the CPU cap when the
cap is lower than the sum of the CPU percentages. So
something like 4 vCPU and a cap of 200 make sense to me.

Can somebody explain me what happens with a setting of
let say 1 vCPU and a cap of 200 ?

In the SDC case what was the idea behind having this kind
of settings ? Does it give any
performance/portability/else improvement ?

I am rethinking my packages and where I previously used
settings like you would use on vSphere, I am no wondering
if I am not doing it totally wrong.


There are 3 important concepts. vCPUs, cpu_shares, and
cpu_cap. From
the vmadm man page (my further comments below that):

   vcpus:

   For KVM VMs this parameter defines the number of
virtual
CPUs the guest
   will see. Generally recommended to be a multiple of 2.

   type: integer (number of CPUs)
   vmtype: KVM
   listable: yes
   create: KVM only
   update: KVM only (requires VM reboot to take effect)
   default: 1

   cpu_shares:

   Sets a limit on the number of fair share scheduler
(FSS)
CPU shares for
   a VM. This value is relative to all other VMs on the
system, so a value
   only has meaning in relation to other VMs. If you
have one VM with a
   a value 10 and another with a value of 50, the VM
with 50 will get 5x
   as much time from the scheduler as the one with 10
when there is
   contention.

   type: integer (number of shares)
   vmtype: OS,KVM
   listabl

[smartos-discuss] Re: Network Problems under Yosemite->VMware Fusion->SmartOS->Zone

[Nahum Shalman]

On 02/16/2016 01:18 AM, 曹锐创 wrote:
I'm trying SmartOS under Yosemite(10.10.1)->VMware 
Fusion(7.0.0)->SmartOS(20160204)->Zone(Minimal Image 14.4.2), but I 
found the network in zones doesn't work. I can't reach the external 
sites so I can't install anything in the zones for experiments. I have 
enable the hypervisor applications and code profile applications under 
VMware, and chose "Shared with my Mac" for Network Adaptor.
Unfortunately I do not have a Mac with VMware Fusion to use to test 
with. I know that the Joyent engineers user VMware Fusion on their Mac 
laptops though.
I'd imagine that you need some sort of bridged networking configuration 
and to enable promiscuous mode in some way.


Google turns up among other things the following two links that may be 
helpful:

https://wiki.smartos.org/display/DOC/SmartOS+as+a+VMware+Guest
https://gist.github.com/jaredmorrow/3053951

I read the doc on 
https://wiki.smartos.org/display/DOC/SmartOS+on+VirtualBox, which said 
the SmartOS works well with Mac + VMware. Could you help me to find 
what's wrong with my settings or forward this mail to jritorto? I 
wonder how the settings is when it works.




I figured that sending this to the mailing list would be even more 
effective. Hopefully anyone who replies will be kind enough to CC you on 
their reply.



Best Regards!


Ruichuang Cao （曹锐创）

Work E-mail:caoruichu...@briphant.com


Good luck!
-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Question regarding package sizing

[Nahum Shalman]

On 02/15/2016 01:30 AM, Benjamin Bergia wrote:

Hi,

I recently noticed that all the packages used by SDC zones are using 
some strange settings. All of them, even database ones, are using 1 
vCPU with a cap of 400. I can picture in my head the "meaning" of the 
CPU cap when the cap is lower than the sum of the CPU percentages. So 
something like 4 vCPU and a cap of 200 make sense to me.


Can somebody explain me what happens with a setting of let say 1 vCPU 
and a cap of 200 ?


In the SDC case what was the idea behind having this kind of settings 
? Does it give any performance/portability/else improvement ?


I am rethinking my packages and where I previously used settings like 
you would use on vSphere, I am no wondering if I am not doing it 
totally wrong.


There are 3 important concepts. vCPUs, cpu_shares, and cpu_cap. From the 
vmadm man page (my further comments below that):


   vcpus:

   For KVM VMs this parameter defines the number of virtual 
CPUs the guest

   will see. Generally recommended to be a multiple of 2.

   type: integer (number of CPUs)
   vmtype: KVM
   listable: yes
   create: KVM only
   update: KVM only (requires VM reboot to take effect)
   default: 1

   cpu_shares:

   Sets a limit on the number of fair share scheduler (FSS) CPU 
shares for
   a VM. This value is relative to all other VMs on the system, 
so a value
   only has meaning in relation to other VMs. If you have one 
VM with a
   a value 10 and another with a value of 50, the VM with 50 
will get 5x

   as much time from the scheduler as the one with 10 when there is
   contention.

   type: integer (number of shares)
   vmtype: OS,KVM
   listable: yes
   create: yes
   update: yes (live update)
   default: 100

   cpu_cap:

   Sets a limit on the amount of CPU time that can be used by a 
VM. The
   unit used is the percentage of a single CPU that can be used 
by the VM.

   Eg. a value of 300 means up to 3 full CPUs.

   type: integer (percentage of single CPUs)
   vmtype: OS,KVM
   listable: yes
   create: yes
   update: yes (live update)


First, note that "vcpus" from a SmartOS perspective only applies to KVM 
VMs. That setting determines how many processors the VM can see and thus 
can use to schedule its processes. We'll come back to that.


Zones (both the kind containing the QEMU process for KVM VMs and regular 
LX and joyent branded ones) can see all of the physical processors and 
the OS can schedule processes on any of them.
If you imagine yourself as a multi-tenant cloud provider you'll quickly 
realize that you need two things:
1. Fairness (preventing noisy neighbors) when the system is fully 
loaded. This is what cpu_shares does. If you give every zone the 
appropriate number of shares they will all get the proportional amount 
of system CPU when the system is fully loaded.
2. Paying for what you get. On a system that is *not* fully loaded, in 
theory a zone could use lots and lots of CPUs. Customers would be 
incentivized  to create and destroy zones until they found one that 
could use lots of free CPU. This is where CPU caps come in. They ensure 
that on a system that is *not* fully loaded the zone can only burst up 
to a reasonable amount relative to what the customer is paying. This 
also helps manage expectations. Setting a CPU cap reasonably close to 
the amount of CPU that the customer gets when the system *is* fully 
loaded means that people are less likely to *think* that they are 
suffering from noisy neighbors (when the delta of how much CPU you get 
on a fully loaded vs fully unloaded system is small, you see more 
consistent performance.)


I haven't looked at the details of the SDC packages, but I can 
confidently say that "vcpu" in the context of a joyent branded zone is 
an approximation of what to expect based on the shares and the cap (as 
opposed to a KVM VM where that's literally the number that the VM will see).


So if you have a SDC service zone with "1 vCPU and a cap of 200" then 
it's getting shares such that when the system is fully loaded it should 
get approximately 1 CPU's worth of CPU time from the scheduler, but when 
the system is not fully loaded it should be able to get up to 2 CPUs 
worth of CPU time from the scheduler but no more. The difference between 
those two is what the Joyent cloud advertises as "bursting"


Coming back for one last moment to KVM VMs, remember that the QEMU 
process is running in a zone that can have shares and caps. 
Additionally, when the VM does I/O, QEMU threads need to be scheduled to 
do some (overhead) work to make that happen.
So in theory you might need your shares and caps to be slightly more 
than just what the number of vCPUs might otherwise suggest (e.g. for 
something performance critical 

Re: [smartos-discuss] Is possible to use images from Docker hub?

[Nahum Shalman]

On 02/05/2016 11:18 AM, Paolo Marcheschi wrote:

Hi

Today i tried to investigate the possibility to use images from 
https://hub.docker.com/explore/

These are thousand of ready made apps.

And it seems that is possible by using the detailed instruction from 
this blog entry:


http://www.smllr.nl/2015/10/11/docker-on-smartos-a-retry/

I  downloaded the busybox and It runs very well

But when i try to  import more complex images like centos or 
wordpress, It downloads a lot of images :


54076c33-de2f-f996-3509-87b662e1ac95  docker-layer  77e39ee82117 
linuxdocker2016-01-25
3b3d6444-6381-bd39-455d-fb39e279eb2a  docker-layer  5eb1402f0414 
linuxdocker2016-01-25
db03997c-ea6f-038f-a367-3c22ffc17138  docker-layer  9287fae7a16e 
linuxdocker2016-01-25
3fc33719-a2f9-0981-f7cf-83340f2d1ad6  docker-layer  e8eb9f547930 
linuxdocker2016-01-26

.

I guess thy are not usable in this form.

Is there a simple way to use them, or I need an SDC-TRITON 
infrastructure to play with them ?


You may be looking for the output of "imgadm list --docker" to help you 
identify the correct uuid...


-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Are there nightly builds for lx-brands? Any chance to access "Hurricane Karen" earlier?

[Nahum Shalman]

On 02/04/2016 06:12 AM, Paul Dunkler wrote:

Hi there,

i'm wondering if there are any "nightly builds" or somewhat for the lx 
brand images.
Is there any way to get an early access to the upcoming release 
"*2016-02-04 Hurricane Karen*(Release Date: 2016-04-02)"?


If i'd like to build it my own, do i need to use 
the centos-lx-brand-image-builder?


Appreciate your help! Thanks in advance.


https://wiki.smartos.org/display/DOC/LX+Branded+Zones#LXBrandedZones-JoyentCIbuilds

Just to clarify, that's an upgrade to the underlying SmartOS platform, 
not a new filesystem of linux binaries.


If you still have questions, let us know.

-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Re: question regarding to file: usb_fdisk_table for smartos

[Nahum Shalman]

Yes, I must have hit reply to the wrong email or something.
I used both of those changes in the image I posted.
One person on IRC reported that image as being slower than the original.
I'd like to track that down before recommending that we upstream these 
changes.


-Nahum

On 01/12/2016 06:28 PM, ZHU XIAN WEN wrote:

Hi Nahum Shalman

I use following disk table to boot usb:

* IdAct  Bhead  Bsect  BcylEhead  Esect  EcylRsect  Numsect
   12128  0  0  0   0  0  0   2563891200
   0 00  0  0   0  0  0   0  0
   0 00  0  0   0  0  0   0  0
   0 00  0  0   0  0  0   0  0

and following #528
The change is here:

https://github.com/joyent/smartos-live/blob/master/tools/build_iso#L187

from:
echo "y" | pfexec mkfs -F pcfs -o fat=32 ${bi_lofidev/lofi/rlofi}:c

to:
echo "y" | pfexec mkfs -F pcfs -o fat=32,spc=32 ${bi_lofidev/lofi/rlofi}:c


then, the performance is significantly faster/better.

Tony


On 1/13/16 06:06, Nahum Shalman wrote:

As a test I've made a USB image using the suggested fdisk table and
also using the change in #528 (referenced below).
It certainly boots when used as a virtual disk in a VM, but if anyone
would like to do some performance testing of it on a real USB stick
and report back, that would be great.

If people are happy with those changes I'll open up a pull request to
get this stuff upstream.

The image is here (built using a CI tarball so use on a test machine
not production):
http://us-east.manta.joyent.com/nahamu/public/smartos/platform-20160112T175701Z-4k-align-16k-cluster.usb.bz2


-Nahum

On 01/11/2016 10:49 AM, Nahum Shalman wrote:

On 01/09/2016 06:21 AM, ZHU XIAN WEN wrote:

hello Nahum Shalman,

this is Tony, I read the disk table of usb_fdisk_table starting from 63
sector, it is not alignment, may I know what's the tool you used to
generate the 2 GB disk table, actually, it is not aligned for 4k,
how to
align it? if I need a 8G usb, how to do it.

* IdAct  Bhead  Bsect  BcylEhead  Esect  Ecyl Rsect
Numsect
12128  9  34 0   34 63 243 600
3905400
0 00  0  0   0  0  0 0  0
0 00  0  0   0  0  0 0  0
0 00  0  0   0  0  0 0  0

thank you very much

best wishes

Hi Tony,

If I recall correctly I think I just created a 2000MB file (slightly
smaller than 2GB to protect against people who might have gotten
"2GB" USB sticks that were slightly smaller than 2048MB) and then
used that as a raw disk for running SmartOS under QEMU.

I think I then just used the standard illumos tools to partition the
disk from within the VM, then dumped out that partition table after
mounting the image as a lofi device.
Unfortunately it's been so long since I did it that I've forgotten
precisely how I did it.

It might be worth revisiting that partition table to align it
properly for 4k block sizes along with testing the change proposed in
https://github.com/joyent/smartos-live/issues/528

I'm hoping others on the mailing list might have some ideas. Ideally
if someone has a physical 2GB device that advertises itself as using
4k blocks, pointing the tools at that device might help generate a
better partition table, etc.

-Nahum








---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Re: question regarding to file: usb_fdisk_table for smartos

[Nahum Shalman]

As a test I've made a USB image using the suggested fdisk table and also 
using the change in #528 (referenced below).
It certainly boots when used as a virtual disk in a VM, but if anyone 
would like to do some performance testing of it on a real USB stick and 
report back, that would be great.


If people are happy with those changes I'll open up a pull request to 
get this stuff upstream.


The image is here (built using a CI tarball so use on a test machine not 
production):

http://us-east.manta.joyent.com/nahamu/public/smartos/platform-20160112T175701Z-4k-align-16k-cluster.usb.bz2

-Nahum

On 01/11/2016 10:49 AM, Nahum Shalman wrote:

On 01/09/2016 06:21 AM, ZHU XIAN WEN wrote:

hello Nahum Shalman,

this is Tony, I read the disk table of usb_fdisk_table starting from 63
sector, it is not alignment, may I know what's the tool you used to
generate the 2 GB disk table, actually, it is not aligned for 4k, how to
align it? if I need a 8G usb, how to do it.

* IdAct  Bhead  Bsect  BcylEhead  Esect  Ecyl Rsect  Numsect
   12128  9  34 0   34 63 243 6003905400
   0 00  0  0   0  0  0 0  0
   0 00  0  0   0  0  0 0  0
   0 00  0  0   0  0  0 0  0

thank you very much

best wishes


Hi Tony,

If I recall correctly I think I just created a 2000MB file (slightly
smaller than 2GB to protect against people who might have gotten "2GB"
USB sticks that were slightly smaller than 2048MB) and then used that
as a raw disk for running SmartOS under QEMU.

I think I then just used the standard illumos tools to partition the
disk from within the VM, then dumped out that partition table after
mounting the image as a lofi device.
Unfortunately it's been so long since I did it that I've forgotten
precisely how I did it.

It might be worth revisiting that partition table to align it properly
for 4k block sizes along with testing the change proposed in
https://github.com/joyent/smartos-live/issues/528

I'm hoping others on the mailing list might have some ideas. Ideally
if someone has a physical 2GB device that advertises itself as using
4k blocks, pointing the tools at that device might help generate a
better partition table, etc.

-Nahum




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

[smartos-discuss] Re: question regarding to file: usb_fdisk_table for smartos

[Nahum Shalman]

On 01/09/2016 06:21 AM, ZHU XIAN WEN wrote:

hello Nahum Shalman,

this is Tony, I read the disk table of usb_fdisk_table starting from 63
sector, it is not alignment, may I know what's the tool you used to
generate the 2 GB disk table, actually, it is not aligned for 4k, how to
align it? if I need a 8G usb, how to do it.

* IdAct  Bhead  Bsect  BcylEhead  Esect  EcylRsect  Numsect
   12128  9  34 0   34 63 243 6003905400
   0 00  0  0   0  0  0   0  0
   0 00  0  0   0  0  0   0  0
   0 00  0  0   0  0  0   0  0

thank you very much

best wishes


Hi Tony,

If I recall correctly I think I just created a 2000MB file (slightly 
smaller than 2GB to protect against people who might have gotten "2GB" 
USB sticks that were slightly smaller than 2048MB) and then used that as 
a raw disk for running SmartOS under QEMU.


I think I then just used the standard illumos tools to partition the 
disk from within the VM, then dumped out that partition table after 
mounting the image as a lofi device.
Unfortunately it's been so long since I did it that I've forgotten 
precisely how I did it.


It might be worth revisiting that partition table to align it properly 
for 4k block sizes along with testing the change proposed in 
https://github.com/joyent/smartos-live/issues/528


I'm hoping others on the mailing list might have some ideas. Ideally if 
someone has a physical 2GB device that advertises itself as using 4k 
blocks, pointing the tools at that device might help generate a better 
partition table, etc.


-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Can't add image source https://datasets.joyent.com/datasets

[Nahum Shalman via smartos-discuss]

On 02/12/2015 10:43 PM, zhiqiang feng via smartos-discuss wrote:
I just install smarts latest version (v0.147+ build: 
20150205T075858Z), and try to add image source described in the user 
guide.


I got this error message:
unexpected ping error with imgapi image source 
https://datasets.joyent.com/datasets: 404Error: Dataset ping not 
found.


the I try to:
imgadm -vvv sources -a https://datasets.joyent.com/datasets


Which user guide were you reading? Was it a wiki page? It probably needs 
updating...


First of all, you should be using https://images.joyent.com rather than 
the older https://datasets.joyent.com/datasets. They contain the same 
images. And the newer one should already be present as a source by default.


But second of all, there was a syntax change for using old DSAPI servers 
when adding them with imgadm. What you would want to do for other DSAPI 
servers is use add -t dsapi to your command.


So what you would have used is:
imgadm sources -a https://datasets.joyent.com/datasets -t dsapi

But again, you probably don't need to do that because you probably 
already have the newer https://images.joyent.com image server already 
configured.


-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] expose socket to non-global zones

[Nahum Shalman via smartos-discuss]

On 02/10/2015 07:48 PM, Marc Phillips via smartos-discuss wrote:

I'm tasked with doing the following:

Expose /some/software/sockets UDS to each of the non-global zones

I have no idea how to do that or what that means, exactly, in the context of 
SmartOS.

I have a PCI card supported by SmartOS which utilies UDS.  Said card and 
supporting software is working
as expected within the global zone.

One thing that would make said software work better would be if I could add a 
generic user/group to the global zone.

Any pointers would be appreciated.

On 02/10/2015 08:51 PM, Garrett D'Amore via smartos-discuss wrote:

Generally, you can create UNIX domain sockets in the global zone, and if you 
put the path to them in a filesystem that is mounted inside the local zone, 
then that local zone will be able to attach.  This is a kind of way to enable 
communication between the GZ and the local zone.

Note that generally zones don’t share filesystems, so typically you can’t use 
this to get between to distinct local zones.  There may be some tricks to make 
even that work, but to be honest I’ve not tried — it seems like it would be a 
bad idea.


I suspect that setting up a lofs mounted filesystem for the zones at 
provision time that mounts in the path where the unix sockets are being 
created would do the trick.


Search the vmadm man page for lofs.

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] LX Branded Zone and updates.joyent.com

[Nahum Shalman via smartos-discuss]

On 02/03/2015 01:35 AM, Greg Zartman via smartos-discuss wrote:
I thought I'd play around with the LX branded zones that the Joyent 
devs have been working hard on.   I was reading this wiki page here:


https://wiki.smartos.org/display/DOC/LX+Branded+Zones#LXBrandedZones-64Bit 



There is a step to import updates.joyent.com 
http://updates.joyent.com as a source of the lx zone image.  When I 
do this and then do a image avail, imgadm just seems to hang.  I was 
hoping to look and see what the latest image happened to be, but I 
can't seem to get a listing.  Is this normal?


Hopefully someone will correct me if I'm wrong...

My understanding is that updates.joyent.com has a *LOT* of images on it, 
so the hang is your machine downloading and parsing a very very large 
amount of json.


I'm fairly certain that there aren't any newer images posted, so it 
should be safe to stick with the image UUIDs on the wiki page.


-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Trimming an SSD drive or migrating the root volume

[Nahum Shalman via smartos-discuss]

On 01/13/2015 06:19 PM, smartos-discuss@lists.smartos.org wrote:

I wonder if it would be possible, after detaching SSD from the mirror in
your step #2, to pass the device over to Linux running locally in KVM and
'secure erase' it from KVM other than physically disconnecting it from the
smartos machine?


Maybe, but I highly doubt it. QEMU doesn't pass through the whole ATA 
device, it passes in a virtual ATA device backed by the blocks that live 
on the real ATA device. It might be interesting to do the experiment, 
though (you'd either have to create a VM with vmadm then muck around 
with zonecfg to pass through the disk, or launch QEMU from the GZ 
manually...)



Which Intel SSD are you using that you haven't had issues with?
DC S3500 or S3700?


These machines are so old that it's the old 320 series ones. From what 
I've read[1], the S3500 is a
decent successor to the 320 series. If you have the money, Joyent use 
the S3700s in some of their BOMs[2].


I doubt that we're getting particularly impressive performance from the 
320s, but they've been reliable and performance has been consistent. The 
machine I ordered most recently will have S3500s, but it hasn't arrived 
yet so I have no data for you...


-Nahum

[1]http://www.tomshardware.com/reviews/ssd-dc-s3500-review-6gbps,3529.html
[2]http://eng.joyent.com/manufacturing/bom.html#tenderloin-c-joyent-compute-platform-2102-2u-32t-256gb-ssd


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Trimming an SSD drive or migrating the root volume

[Nahum Shalman via smartos-discuss]

On 01/12/2015 06:56 PM, smartos-discuss@lists.smartos.org wrote:

I have the root volume with several I/O intensive zones on an SSD drive.
Updating a file in one of the zones is noticeably going slow in comparison
how it was several months ago, while the file size hasn't changed much. I
suspect the SSD may need trimming.
Is the TRIM not available yet in smartos?
If so, what would be the right procedure of migrating the root volume with
its zones/* onto the different drive?


This advice may not help you since it sounds like you have your zones 
pool on a single SSD rather than a mirrored pair, but perhaps it will 
help someone with a similar problem.


We have some SmartOS machines with Intel SSDs, and some with Samsung 
SSDs (I think it's the 840Pro ones, but I'm not certain). We haven't had 
issues with the machines with the Intel SSDs, but we started seeing slow 
I/O on some of the machines with the Samsungs.


As you guessed, illumos lacks trim support. Since I have a mirrored pool 
I was able to do the following:


1. Offline one of the disks.
2. Pull it out, put it in a Linux machine and do this: 
https://ata.wiki.kernel.org/index.php/ATA_Secure_Erase

3. Put it back in the SmartOS machine and put it back into the mirror
4. Wait for the mirror to resilver
5. Repeat for the other disk.

Performance was restored, though I anticipate the problem will recur at 
some point. The long term plan is to replace those SSDs with Intel ones 
that don't seem to degrade over time.


It would be nice to be able to issue that ATA command to the SSDs 
directly from illumos, but at the moment I don't know how and the last 
time I asked I didn't get very far.


So ideally, I would recommend buying a replacement SSD whose performance 
won't degrade, attach it to the current drive to make a mirror, then 
remove your current SSD from the pool and use it in a laptop with an OS 
that supports TRIM.


Good luck!

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

[smartos-discuss] Better serial port support for the SmartOS GRUB menu

[Nahum Shalman via smartos-discuss]

TL;DR? If/when smartos-live#379 gets applied, the mechanism for choosing 
ttya as the system console when booting SmartOS will change to match how 
it's done in SDC. You have all been warned. :)


Still here?

Now that Keith has fixed up GRUB[1] I've written some code[2] that would 
let SmartOS use two features currently only available on the SDC USB sticks:


1. GRUB renders on both the VGA output and the serial port.
2. There's an os_console variable that lets you switch where illumos 
should put the console messages (vga vs ttya, ttyb, etc) rather than 
having lots of different GRUB entries.


In addition, If you build SmartOS yourself, you would now be able to 
specify which console SmartOS should use by default when you're building 
your own media.


If you build SmartOS yourself, I invite you to try out this new GRUB 
menu and put any feedback in the pull request.


Many thanks to Keith for fixing those GRUB issues.

Thanks for reading!
-Nahum

[1]
 * 
https://github.com/joyent/illumos-joyent/commit/7700606bb48dbeb21848ca88265e7e5fac5b174f

 * https://smartos.org/bugview/OS-3634
 * https://smartos.org/bugview/OS-3635

[2] https://github.com/joyent/smartos-live/pull/379



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

[smartos-discuss] Fixing SmartOS USB template (was Re: SmartOS on a Intel NUC - USB Keyboard not working after boot)

[Nahum Shalman via smartos-discuss]

On 11/25/2014 11:36 AM, Keith Wesolowski via smartos-discuss wrote:

That's the interesting part of the job, for sure.  The uninteresting
part is replacing the horrible bits in Jenkins that we use to build
this thing with something that looks like what we use to build
everything else.  See

/wesolows/public/build-smartos - the way Jenkins invokes this pile
/wesolows/public/jenkins-tmpl - the way Jenkins invokes normal MG jobs

Then look at
https://github.com/joyent/mountain-gorilla/blob/master/tools/jenkins-build
and figure out how to get MG to build SmartOS properly.  Once you've
done that, it should be simple to see how it all fits.  With the change
to MG in place (and any needed for smartos-live etc. as well), and new
template images, we can switch over Jenkins.

The next step is to ditch these template images entirely and create them
on the fly.  That will require some kernel work in lofi, or some better
tools for creating fake geometry and fake partition tables on loopback
files.


All necessary changes to smartos-live and mountain-gorilla have now 
landed [1].
With some additional QA on Joyent's end, you should be able to switch 
over to Jenkins now.
The MG target is smartos and if it detects that it is building a 
release branch it will automatically create appropriate release 
snaplinks as well.


Many thanks to Keith, Robert, and Trent for their help.

-Nahum

[1]
https://github.com/joyent/smartos-live/pull/367
https://github.com/joyent/smartos-live/pull/368
https://github.com/joyent/mountain-gorilla/pull/2
https://github.com/joyent/mountain-gorilla/pull/3


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] joyent_20141016T192648Z

[Nahum Shalman via smartos-discuss]

How big is your swap zvol?
Did you upgrade the RAM in that machine some time after the initial 
installation of SmartOS?

If so you might need to grow your swap zvol...

-Nahum

On 11/12/2014 04:56 PM, David Kelbley via smartos-discuss wrote:

Has anyone else seen a problem with joyent_20141016T192648Z pertaining to 
memory allocation errors.  When attempting to start a KVM, I receive the 
following:

Failed to allocate memory: Resource temporarily unavailable

Yet, vmstat shows the following:

kthr  memorypagedisk  faults  cpu
  r b w   swap  free  re  mf pi po fr de sr lf rm s0 s1   in   sy   cs us sy id
  0 0 22 155883264 42442864 232 2652 13 1109 1113 165056 22459 1 -1556 61 -0 
46825 394574 157957 1 12 87

The allocated memory to the KVM is only 4GB where as I have 42GB free.   Any 
thoughts on what's happening?

Thanks,
Dave




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/22280019-340ab187
Modify Your Subscription: https://www.listbox.com/member/?;
Powered by Listbox: http://www.listbox.com




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] SmartOS on Supermicro X9SBAA-F with Marvell 88SE9230 SATA3

[Nahum Shalman via smartos-discuss]

On 11/03/2014 05:29 AM, Christopher J. Ruwe via smartos-discuss wrote:

Hello,

I am trying to run SmartOS on a Supermicro Atom S1260 board. I have
circumvented the USB 3.0 problem using the serial console over
ethernet feature (IMPI/SoL). Trying to install, however, fails,
because all connected disks are invisible on the  USB-booted machine.

The board in question
(http://www.supermicro.com/products/motherboard/ATOM/X9/X9SBAA-F.cfm)
has a Marvell 88SE9230 SATA3 controller which I suspect to be the
problem. The SATA controller works fine on pfSense 2.1 (FreeBSD 8.3)
and later FreeBSD flavours.

Can someone verify my suspicion or does somebody have an idea how to
get SmartOS working on that machine?


Unless something has changed your experience matches my current 
understanding that USB3 is broken in illumos.
If your BIOS allows you to disable the USB3 controller, you should be 
able to use the machine.


If I'm horribly mistaken, hopefully someone with better information will 
correct me.


-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] setting ngroups_max in zones

[Nahum Shalman via smartos-discuss]

On 10/14/2014 03:01 AM, Ian Collins via smartos-discuss wrote:
I believe ngroups_max has to be set too soon for 3 and I don't think 
it is editable via mdb.  I've seen Keith's blog and I'm investigating 
2 on a test box as a long term solution.


I wrote up a concrete example that may be helpful:
http://blog.shalman.org/overriding-driver-config-files-on-smartos/

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] External interface as physical nic in GZ, not using a vnic

[Nahum Shalman via smartos-discuss]

Your external_nic line is correct.
The other three should be named like so:
external0_ip
external0_netmask
external0_gateway

-Nahum

On 10/07/2014 07:26 AM, Christian Meunier via smartos-discuss wrote:

Hi all,

I am new to smartos and I am trying to install it on one server which 
has 2 nics (ixgbe0 for external and ixgbe1 for admin).


The admin interface is correcly setup and ifconfig shows that the 
ixgbe1 is up with the desired network config. However, for the 
external, instead of bringing up the ixgbe0 with the desired ip, the 
system has created a vnic on top of it and brough it up.


How can I tell smartos that I dont want to create a vnic for the 
external interface but I want to use the ixgbe0 directly. Basically, 
having the same behaviour as to how the admin nic is setup.


The config for the external nic I use is as follow :

external_nic={mac-address-omitted}
external_ip={public-ip-omitted}
external_netmask=255.255.255.0
external_gateway={public-ip-omitted}

beside these line and the ones to config the admin nic, i have only:

- headnode_default_gateway
- dns_resolvers
- dns_domain
- ntp_hosts

Thanks in advance,

--
Chris
*smartos-discuss* | Archives 
https://www.listbox.com/member/archive/184463/=now 
https://www.listbox.com/member/archive/rss/184463/22280019-340ab187 
| Modify 
https://www.listbox.com/member/?; 
Your Subscription 	[Powered by Listbox] http://www.listbox.com







---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Snapshots

[Nahum Shalman via smartos-discuss]

On 09/23/2014 07:15 PM, Khushil Dep via smartos-discuss (Personal) wrote:

It’s not the same file set each time. I’m trying to find an elegant solution to 
snapshotting files on change.


What's changing the files? If it's something like cfengine, shouldn't 
that tool provide a means to back up the file it's modifying? (I'm 
pretty sure that's what puppet does...)


Ideally the tool that's doing the modifying should store off a copy of 
the old file.
You could use something like etckeeper[1] to track changing files in git 
(or just use git directly)


But ultimately you either need cooperation from the tool that is 
changing the files, or you need to be regularly snapshotting the file 
system, monitoring all the files you care about for changes, and when a 
file gets changed you can copy out the previous version from a snapshot 
into the place where you want to keep the old versions...


Or maybe there's a better way someone smarter than I am will suggest...

-Nahum

[1] https://joeyh.name/code/etckeeper/



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Snapshots

[Nahum Shalman via smartos-discuss]

I think you should be able to write a user-space application using event 
ports to watch the files for changes. If you've been snapshotting the 
filesystem regularly, you can reach into .zfs/snapshot/snapname and 
copy out the previous version (this of course assumes delegated zfs...)


Otherwise the other ugly user-space hack that comes to mind is you use 
something like node-nfs to write a little server that keeps old versions 
around behind the scenes, run that, and mount it up via nfs...


-Nahum

On 09/24/2014 10:58 AM, GMail via smartos-discuss wrote:

If only this was CFengine! I would have a really elegant solution to my 
problem. Many programs could change this file. I think im off to write 
something in kernel maybe Oh good.

Sent from my iPhone


On 24 Sep 2014, at 15:43, Nahum Shalman nshalman-smar...@elys.com wrote:


On 09/23/2014 07:15 PM, Khushil Dep via smartos-discuss (Personal) wrote:
It’s not the same file set each time. I’m trying to find an elegant solution to 
snapshotting files on change.

What's changing the files? If it's something like cfengine, shouldn't that tool 
provide a means to back up the file it's modifying? (I'm pretty sure that's 
what puppet does...)

Ideally the tool that's doing the modifying should store off a copy of the old 
file.
You could use something like etckeeper[1] to track changing files in git (or 
just use git directly)

But ultimately you either need cooperation from the tool that is changing the 
files, or you need to be regularly snapshotting the file system, monitoring all 
the files you care about for changes, and when a file gets changed you can copy 
out the previous version from a snapshot into the place where you want to keep 
the old versions...

Or maybe there's a better way someone smarter than I am will suggest...

-Nahum

[1] https://joeyh.name/code/etckeeper/



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/22280019-340ab187
Modify Your Subscription: https://www.listbox.com/member/?;
Powered by Listbox: http://www.listbox.com




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] nfs/status and nlockmgr not starting after upgrade, NFS also not running

[Nahum Shalman via smartos-discuss]

My home machine serves NFS.
I'm currently running the latest platform (joyent_20140821T185718Z) and 
everything is fine.


# svcs | grep nfs
online Aug_25   svc:/network/nfs/status:default
online Aug_25   svc:/network/nfs/nlockmgr:default
online Aug_25   svc:/network/nfs/mapid:default
online Aug_25   svc:/network/nfs/server:default

So either there was a bug that isn't in the current platform, or there's 
a misconfiguration on your machine.


-Nahum

On 08/26/2014 12:24 PM, Sam M via smartos-discuss wrote:

Hello all.

After the latest upgrade of the SmartOS platform, exactly a month ago, 
I'm unable to get NFS to start. Nothing in the logs either. Anyone 
seeing anything similar? Is there a fix for this?


TIA.

Sam


[root@sequoia ~]# uname -a
SunOS sequoia 5.11 joyent_20140807T193112Z i86pc i386 i86pc

[root@sequoia ~]# svcs -xv
svc:/network/nfs/status:default (NFS status monitor)
 State: maintenance since 26 August 2014 16:20:48 UTC
Reason: Restarting too quickly.
   See: http://illumos.org/msg/SMF-8000-L5
   See: man -M /usr/share/man -s 1M statd
   See: /var/svc/log/network-nfs-status:default.log
Impact: 3 dependent services are not running:
svc:/network/nfs/nlockmgr:default
svc:/network/nfs/client:default

*smartos-discuss* | Archives 
https://www.listbox.com/member/archive/184463/=now 
https://www.listbox.com/member/archive/rss/184463/22280019-340ab187 
| Modify 
https://www.listbox.com/member/?; 
Your Subscription 	[Powered by Listbox] http://www.listbox.com







---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Current Standalone SaltStack Builds on Joyent Base64 image version 13.4.2

[Nahum Shalman via smartos-discuss]

Yikes.

You shouldn't have to install any dependencies. I must have broken 
something...


I'll see what I can figure out and post back soon.

-Nahum


On 07/14/2014 12:32 PM, Michael Noack via smartos-discuss wrote:

Has someone been able to run:
- http://www.shalman.org/salt/salt-2014.1.6-esky-smartos.tar.gz
or
- http://www.shalman.org/salt/salt-2014.1.7-esky-smartos.tar.gz
on SmartOS?

I've been trying on Joyents Base64 image version 13.4.2 with the zeromq 
package installed (zeromq-3.2.4). All prior versions up to 
http://www.shalman.org/salt/salt-2014.1.5-esky-smartos.tar.gz run just fine. I suspect a 
dependency that I might be missing but I wasn't able to deduct it from the error message.

This is what I did:

 pkgin -y install zeromq
 cd /opt
 curl -kLO http://www.shalman.org/salt/salt-2014.1.6-esky-smartos.tar.gz
 tar xzvf salt-2014.1.6-esky-smartos.tar.gz
 /opt/salt/install/install.sh

When enabling the service, dmesg gets spammed by this message:
2014-07-14T16:26:38+00:00 localhost svc.startd[66286]: [ID 694882 
daemon.notice] instance svc:/network/salt-minion:default exited with status 1

When running the minion manually, I get the following error:

# /opt/salt/bin/appdata/salt-2014.1.6.solaris-2_11-i86pc_64bit/salt-minion
[CRITICAL] Failed to load grains defined in grain file core.zmqversion in function 
function zmqversion at 0x14e21b8, error:
Traceback (most recent call last):
   File salt/loader.py, line 1034, in gen_grains
   File salt/grains/core.py, line 1205, in zmqversion
   File zmq/__init__.py, line 62, in module
   File zmq/backend/__init__.py, line 22, in module
   File zmq/backend/select.py, line 31, in select_backend
ImportError: No module named cffi
[ERROR   ] An un-handled exception was caught by salt's global exception 
handler:
NameError: global name 'zmq' is not defined
Traceback (most recent call last):
   File string, line 6, in module
   File __main__.py, line 128, in module
   File __main__salt-minion__.py, line 17, in module
   File salt/scripts.py, line 35, in salt_minion
   File salt/__init__.py, line 221, in start
   File salt/__init__.py, line 209, in prepare
   File salt/minion.py, line 548, in __init__
   File salt/minion.py, line 1073, in authenticate
   File salt/crypt.py, line 359, in sign_in
   File salt/payload.py, line 166, in __init__
NameError: global name 'zmq' is not defined
Traceback (most recent call last):
   File string, line 6, in module
   File __main__.py, line 128, in module
   File __main__salt-minion__.py, line 17, in module
   File salt/scripts.py, line 35, in salt_minion
   File salt/__init__.py, line 221, in start
   File salt/__init__.py, line 209, in prepare
   File salt/minion.py, line 548, in __init__
   File salt/minion.py, line 1073, in authenticate
   File salt/crypt.py, line 359, in sign_in
   File salt/payload.py, line 166, in __init__
NameError: global name 'zmq' is not defined
Exception AttributeError: 'SREQ' object has no attribute 'poller' in bound method 
SREQ.__del__ of salt.payload.SREQ object at 0x14e81d0 ignored




Thanks a lot for your time,
Michael Noack


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/22280019-340ab187
Modify Your Subscription: https://www.listbox.com/member/?;
Powered by Listbox: http://www.listbox.com




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Current Standalone SaltStack Builds on Joyent Base64 image version 13.4.2

[Nahum Shalman via smartos-discuss]

2014.1.6 had problems upstream and I've deleted my build of it

The 2014.1.7 build had an issue (details below). I have rebuilt it and 
uploaded it (in place... just need to redownload).

If there are any issues with the fresh build please let me know.

Details:

All of my previous builds were done in a 2013Q4 pkgsrc environment.

These two were built in a 2014Q1 environment with a newer 
py27-zmq-14.0.1.tgz package which seems to work differently internally 
than py27-zmq-13.x did.


I tried a quick hack to see if I could get things to play nicely with 
pyzmq-14 but it didn't work.


As such I've rebuilt 2014.1.7 in the same build environment as the 
previous builds.


Sorry for the inconvenience!!

-Nahum

On 07/14/2014 12:32 PM, Michael Noack via smartos-discuss wrote:

Has someone been able to run:
- http://www.shalman.org/salt/salt-2014.1.6-esky-smartos.tar.gz
or
- http://www.shalman.org/salt/salt-2014.1.7-esky-smartos.tar.gz
on SmartOS?

I've been trying on Joyents Base64 image version 13.4.2 with the zeromq 
package installed (zeromq-3.2.4). All prior versions up to 
http://www.shalman.org/salt/salt-2014.1.5-esky-smartos.tar.gz run just fine. I suspect a 
dependency that I might be missing but I wasn't able to deduct it from the error message.

This is what I did:

 pkgin -y install zeromq
 cd /opt
 curl -kLO http://www.shalman.org/salt/salt-2014.1.6-esky-smartos.tar.gz
 tar xzvf salt-2014.1.6-esky-smartos.tar.gz
 /opt/salt/install/install.sh

When enabling the service, dmesg gets spammed by this message:
2014-07-14T16:26:38+00:00 localhost svc.startd[66286]: [ID 694882 
daemon.notice] instance svc:/network/salt-minion:default exited with status 1

When running the minion manually, I get the following error:

# /opt/salt/bin/appdata/salt-2014.1.6.solaris-2_11-i86pc_64bit/salt-minion
[CRITICAL] Failed to load grains defined in grain file core.zmqversion in function 
function zmqversion at 0x14e21b8, error:
Traceback (most recent call last):
   File salt/loader.py, line 1034, in gen_grains
   File salt/grains/core.py, line 1205, in zmqversion
   File zmq/__init__.py, line 62, in module
   File zmq/backend/__init__.py, line 22, in module
   File zmq/backend/select.py, line 31, in select_backend
ImportError: No module named cffi
[ERROR   ] An un-handled exception was caught by salt's global exception 
handler:
NameError: global name 'zmq' is not defined
Traceback (most recent call last):
   File string, line 6, in module
   File __main__.py, line 128, in module
   File __main__salt-minion__.py, line 17, in module
   File salt/scripts.py, line 35, in salt_minion
   File salt/__init__.py, line 221, in start
   File salt/__init__.py, line 209, in prepare
   File salt/minion.py, line 548, in __init__
   File salt/minion.py, line 1073, in authenticate
   File salt/crypt.py, line 359, in sign_in
   File salt/payload.py, line 166, in __init__
NameError: global name 'zmq' is not defined
Traceback (most recent call last):
   File string, line 6, in module
   File __main__.py, line 128, in module
   File __main__salt-minion__.py, line 17, in module
   File salt/scripts.py, line 35, in salt_minion
   File salt/__init__.py, line 221, in start
   File salt/__init__.py, line 209, in prepare
   File salt/minion.py, line 548, in __init__
   File salt/minion.py, line 1073, in authenticate
   File salt/crypt.py, line 359, in sign_in
   File salt/payload.py, line 166, in __init__
NameError: global name 'zmq' is not defined
Exception AttributeError: 'SREQ' object has no attribute 'poller' in bound method 
SREQ.__del__ of salt.payload.SREQ object at 0x14e81d0 ignored




Thanks a lot for your time,
Michael Noack


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/22280019-340ab187
Modify Your Subscription: https://www.listbox.com/member/?;
Powered by Listbox: http://www.listbox.com




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] ghc ./configure generating bad config.status:

[Nahum Shalman via smartos-discuss]

I've had issues from time caused by the fact that /bin/sh is a symlink 
to ksh93 and not bash.


There's sadly a large amount of code out there that puts #!/bin/sh at 
the top but actually uses bash-isms...


-Nahum

On 07/10/2014 09:16 PM, Alain O'Dea via smartos-discuss wrote:

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Something seems wrong with autoreconf or the version /bin/sh on SmartOS.

config.status should contain:

test -n $AWK || AWK=awk
# The default lists apply if the user does not specify any file.

It does contain:

test -n $AWK || AWK=awk# The default lists apply if the user does
not specify any file.

If I run ./configure I get the problem above.  If I run bash
./configure it runs without issues.

This does not occur on Solaris x86, FreeBSD amd64, OS X x86_64, or
Linux on any platform:
http://haskell.inf.elte.hu/builders/

I can work around this, but it concerning.

Any ideas?
-BEGIN PGP SIGNATURE-
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTvzreAAoJEP0rIXJNjNSASukH/1+PDF6Jv326FHD53eLWzmXv
8AH7fMarYIhwKMW2eJT/IWJ9aHrR6gIn3OpkqwRC+BanPaY17S5vxPRtVBlVuLSJ
GTPkWdsbELLqBm9WvEtMFHPOt3qzrBH4Xji45MdG3b7b+0Ego+4EtdqyW92NjMBS
xUm8r91ozBwqIHQB0kJ6Y8Vaqjh1bAP6+RRLgVhSfufAmLeilZzQ+qllrDFOW2uj
egXx4yYREGv77x0SMVRlmr599HPkeEeMnymlyuatPp4CotH84Vn3vx63Q8SeP0lu
WMng/X3C7LNOrvEVN3g7Y1M8Qn4GUJGjkosDOVwbTsTQU9aJtlN/zuatDTZLljk=
=/CNY
-END PGP SIGNATURE-


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/22280019-340ab187
Modify Your Subscription: https://www.listbox.com/member/?;
Powered by Listbox: http://www.listbox.com




---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] How to persistentli set guest network bandwidth limit

[Nahum Shalman via smartos-discuss]

On 06/12/2014 04:52 AM, Tim Boudreau via smartos-discuss wrote:
A bit of fun with vmadm list | cut -b-36 | xargs ... ought to let you 
enumerate all the zones and call dladm on each.


May I suggest the cleaner:
vmadm list -H -o uuid | xargs ...

The vmadm command is quite capable of listing only the details you want. :)

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

[smartos-discuss] Re: virtio-net (was Re: pkgsrc update to latest release)

[Nahum Shalman via smartos-discuss]

On 06/11/2014 10:22 PM, Robert Mustacchi via smartos-discuss wrote:

While there isn't currently a virtio-net driver for illumos, this works
fine otherwise. I use illumos in kvm quite frequently for testing.


Nexenta have a pair of patches in their master branch that they ship for 
the vioif driver.
Would it make sense to have some people test drive it in SmartOS before 
trying to get it upstream?


https://github.com/Nexenta/illumos-nexenta/commits/master/usr/src/uts/common/io/vioif

I don't know why no one has gotten it upstream yet.

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] pkgsrc update to latest release

[Nahum Shalman via smartos-discuss]

On 06/12/2014 06:21 AM, Filip Hajny via smartos-discuss wrote:

12. 6. 2014 v 10:45, Tim Boudreau via smartos-discuss 
smartos-discuss@lists.smartos.org:


Someone can correct me if this is likely to have unanticipated side effects, 
other than the ones already mentioned in this thread (i.e. you need to be 
*really* confident that the packages you update will update cleanly - and do a 
snapshot first!).

Yes, definitely make a snapshot before you start, and if the VM provides any public 
service and you can’t just take it out of a balanced pool, make sure you handle it 
as a planned downtime. Of course, ideally you’d have your entire VM scripted up in 
Chef or similar (heck, even Bash scripts are better than nothing), but sometimes 
that’s just sometimes not possible. (One those I’ll just handle this one 
update manually and I swear I’ll script up the setup next time” things.)

For the records again (since I write from a Joyent address), Joyent doesn’t 
recommend that you do this.

I’ve done this myself way too many times (not recently though) and we even 
tried to come up with a script that takes the user through known obstacles. 
However, it turned out there’s just no way to prepare for all possible 
consequences and factors, and providing a script immediately makes this a 
supported procedure in many eyes, so…

In general, issues are less likely to crop up if you’re only moving by a 
quarter, compared to two years, and if you have a minimum number of packages 
(e.g. started from ‘base’) rather than way too many (started from ‘standard’ or 
even the old ‘smartosplus’).

It’s crucial to check the initial “I’m going to upgrade all these packages for 
you” message that pkgin gives you (and yes, it’s annoying difficult to read and 
parse) and look for any major upgrades of MySQL or PostgreSQL. Especially 
PostgreSQL updates (e.g. moving from 9.1 to 9.2) are a pain, because you need 
to create a full dump of your storage before you even start the update, then 
import it into a blank one again. This may just apply to any package that 
stores user data in its own format.

Another typical problem you may hit is when a config file changes significantly 
or even stops working because of a major update. One of the good things about 
pkgsrc is that every single file that a package installs under $PREFIX/etc is 
copied there at installation time from an example under 
$PREFIX/share/examples/$package. So what you could do diff each config file 
against the *old* example file and save that as a patch. If you get stuck on 
the updated VM later on, you could take the *new* example config file, apply 
the patch to it (or transplant the changes manually) and fix the problem.


Another of the issues I ran into is when packages *don't* get new 
version numbers in the new quarter, things can get weird.
I wanted to remove the gcc47-runtime package from my zone transitioning 
from 2013Q4 to 2014Q1 and discovered a bunch of packages were still 
depending on it. Why? Because they had identical version numbers in the 
new quarter. I had to forcibly reinstall all of them before I could 
remove the obsolete package.


You can get yourself into a very messy state that mostly works, but will 
be painful for the next upgrade.
(Or at the very least is wasteful of memory because you have two 
different versions of libgcc being used side by side... no big deal in a 
big zone, but in a busy zone with a small memory cap it can all start to 
add up...)


So yes, if you don't mind the bumps and bruises, snapshot the zone, and 
go for it.


Just keep in mind what Jonathan wrote in his email.
There's a spectrum between we'll take care of all the upgrade details 
for you and here's the latest and greatest, deploy at will.

Ubuntu is the epitome of the former, and pkgsrc, of the latter.

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] pkgsrc update to latest release

[Nahum Shalman via smartos-discuss]

On 06/11/2014 02:55 PM, Alain O'Dea via smartos-discuss wrote:

It is probably possible, but with sharp edges.


The edges are very sharp, even in the simple case.
Having just done it last week I would summarize with:
If you have to ask how to do it, you shouldn't.

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Standalone SaltStack Builds (Re: smartos/ansible?)

[Nahum Shalman via smartos-discuss]

On 05/27/2014 08:03 PM, Dewey Hylton wrote:
agreed - perfectly on-topic as far as i'm concerned. i currently have 
zero invested in any of these cfg management tools, so i'm game to 
try. ansible interested me because of its python backend, and because 
it's agent-less (less going on, less to learn, and less to deal with 
if i don't like ansible for whatever reason). salt at least fits one 
of those.


smartos isn't the only platform i need to support, though ... i'm also 
currently managing solaris 10, suse linux, redhat linux, openbsd, and 
freebsd. any feel for how well salt (and/or ansible) can manage all 
those platforms?


This is starting to veer off-topic for SmartOS, but I do have two thoughts:

1. As Mono said, Salt is great for Linux. That's the environment where 
it sees the most testing and use. The Solaris support predates 
improvements specifically targeted at SmartOS, so one would hope it's 
pretty good (also, some of the other improvements I've personally 
contributed to Salt, e.g. in the realm of SMF support (which sadly still 
could use a bit more work...) apply to Solaris as well). I don't know 
about FreeBSD.


2. In theory, if the build process I run was moved to a Solaris 10 host 
(a S10 branded zone would probably work too), as long as it would be 
possible to get the desired version of Python and libraries installed, 
the resulting build should theoretically run fine on all descendants of 
SunOS 5.10 (SmartOS, OmniOS (which I'm told was able to use the existing 
builds), other illumos distributions, and for those who have to run it, 
Solaris 10 and up.) If there's demand for that, people should contact me 
off-list and/or on IRC in #salt...


I haven't used Ansible, but I would hope that it would be happy as long 
as you had Python on the client machines.


-Nahum



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] DHCP for zones

[Nahum Shalman]

On 05/04/2014 12:23 PM, Robert Mustacchi wrote:

The dhcp option in vmadm today does not, to my limited knowledge, ask
dhcp for resolvers. Though someone should look at snoop output for dhcp
and verify what it is and what it isn't sending.


It's even easier than using snoop. The dhcp client still knows what you 
received.

see dhcpinfo(1) ( online at http://illumos.org/man/1/dhcpinfo )

Basically you'll want to set up something that at zone boot time does 
something sort of like:
echo -e domain $(/sbin/dhcpinfo 15)\n$(/sbin/dhcpinfo 6| sed 
's/^/nameserver /')  /etc/resolv.conf



While you can use the resolvers and maintain_resolvers properties that
you mention, that's obviously not the best way. Looking into this is
pretty low on our priority list because we actually never use dhcp at
Joyent. If folks could dig into this and figure out whether or not we're
actually requesting DNS servers or not or if it's being clobbered for
some reason, than we can go from there.


I theory one could imagine a vmadm field that says to use the resolvers 
from the dhcp server
that would alter the zone behavior, but as it's low on Joyent's 
priorities, you might have to write it yourself

or find someone in the community to write it and get it upstream.

-Nahum


---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Re: [smartos-discuss] Heartbleed OpenSSL vulnerability

[Nahum Shalman]

To be clear, I responded quickly and vocally only in case someone who 
does run something

that would break would see it before actually breaking things horribly.

For many people that solution can work.

I've been regularly doing that on a zone I run in the JPC
that only really runs nginx and some node.js applications.
It started at 2012Q1 and is now on 2013Q4.

But that zone has no database and before doing those upgrades I always 
back everything up

in a way that makes it easy to roll back if the upgrade doesn't go smoothly.

So, to second Renato's suggestion:
*IF* you don't have any applications that will break horribly,
and if you have good backups and are confident you could recover from pkgsrc
screwing up everything royally, then updating all your packages to the 
latest pkgsrc

release is an option to consider.

Renato, I hope you can forgive my tone. I only wanted to protect others from
possibly breaking databases and becoming sad and/or angry.

-Nahum

On 04/10/2014 10:47 AM, renato morano wrote:

sorry :(

On 10 April 2014 16:46, renato morano renato.mor...@gmail.com wrote:

I complain



On 10 April 2014 16:44, Nahum Shalman nshalman-smar...@elys.com wrote:

BE CAREFUL WITH THAT!!!

That is not necessarily safe if your zone runs, for example, Postgres.
Postgres minor version
upgrades break on-disk format compatibilty and pkgsrc doesn't do anything to
handle that
if you just switch which branch you're on like that.

For some zones that's okay, but I would never make that as a blanket
recommendation!

-Nahum


On 04/10/2014 10:42 AM, renato morano wrote:

vi /opt/local/etc/pkgin/repositories.conf

then

from  http://pkgsrc.joyent.com/packages/SmartOS/^^2013Q1^^/i386/All

to

http://pkgsrc.joyent.com/packages/SmartOS/^^2013Q4^^/i386/All

you can find more information about pkgsrc on
http://pkgsrc.joyent.com/installing.html

:D


On 10 April 2014 16:30, Matthew Law m...@webcontracts.co.uk wrote:

On Thu, April 10, 2014 11:38 am, Filip Hajny wrote:

This part suggests that nothing happened at all - pkgin should have
realized that the upstream database changed, and update the local one
too.
Can you try 'pkgin -y -f up' to force the update?

Worst case, you can always 'pkg_add -u openssl' using the core canonical
pkg_add tool.

[root@myzone ~]# pkgin -y -f up
database for http://pkgsrc.joyent.com/packages/SmartOS/2013Q1/i386/All is
up-to-date

[root@myzone ~]# pkgin -y -f install openssl
calculating dependencies... done.
nothing to do.

I'm stumped by this.  Any ideas? - I'd rather stick to the high level
tools if possible rather than using pkg_add, etc.

Thanks,

M.



---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed:
https://www.listbox.com/member/archive/rss/184463/22541654-e1fbdf62
Modify Your Subscription: https://www.listbox.com/member/?;

Powered by Listbox: http://www.listbox.com






---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed:
https://www.listbox.com/member/archive/rss/184463/22541654-e1fbdf62
Modify Your Subscription:
https://www.listbox.com/member/?;
Powered by Listbox: http://www.listbox.com



--
Free as in free speech, not as in free beer.   (RMS)
The more you do the more you own. (AL)
Run zpool scrub periodically and sleep more soundly

Tutto ciò che si dona, ritorna...
Renato Morano
cell.3284117069







---
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com