D (process context
identifier) support
This release also includes experimental support for bhyve branded zones.
# General Info
Every second Thursday we roll a "release-MMDD" release branch and
builds for SmartOS (and Triton DataCenter and Manta, as well).
Cheers,
Josh Wilsdon, on b
>
>
> on December 11th I posted a new problem report to this list as a reply in
> a thread of 2016. Has anybody an idea how I can tackle this problem?
>
> vmadm create takes minutes to provision an 8 GiB /vserver/ before it times
> out:
>
>timed out waiting for
>/var/svc/provisioning to mov
>
>
> zonecfg:c85ac5c4-7eda-4a95-8e6a-d69df52457f8:fs> set type=lfos
>
Hi Chad,
I think the problem here is that you spelled "lofs" wrong. Could you try
again with "lofs" instead of "lfos"?
Thanks,
Josh
---
smartos-discuss
Archives: https://www.listbox
>
> Is there any plan form Joyent to update this Library to a newer node
> version? This is an really old version and besides the security point many
> commands like "const" are not supported :/
>
Hi Kilian,
We would like to eventually update the node in the platform, but this is
not a current pr
>
>
> The reason why I haven't got the concerning entry is lines from #36 to
> #38 of the source (https://github.com/joyent/smartos-live/blob/
> 64998c0b5a7c33287eb7dd118fef4aea42e5a6f3/src/vm/sbin/rotate-
> kvm-logs.sh#L36) --
>
> "
> if [[ ! -s ${logfile} ]]; then
> # file doesn't exist
>
>
> Yes. But "/usr/vm/sbin/rotate-kvm-logs.sh script" won't add entries in
> "/var/logadm/timestamps" like following:
>
"/zones/c3f09892-f36e-4784-9853-b4ceb3e762da/root/tmp/vm.log -P 'Wed Jan 4
> 05:20:15 2017'"
>
> The weirdness is how this type of entry has been added per every kvm
> instan
> But there is nothing about rotate-kvm-log in /etc/logadm.conf from stock.
> It is weird. How were these entries had been added in time stamp? It should
> be logadm command invoking. But where was it invoked? I haven't done it
> manually.
>
>
It doesn't go in logadm.conf. The rotation happens via
>
>
> Thanks for the support, it seems to work. We have running that patched
> VM.js
> now for five days and the zonevent processes are gone.
>
>
Glad to hear it Stefan!
Thanks,
Josh
---
smartos-discuss
Archives: https://www.listbox.com/member/archive/18
>
>
> I am running joyent_20161013T025521Z now. It seems in this version
> rotate-kvm-log is done by cron not logadm.
> But I also noticed when a kvm instance was created there was a
> corresponding entry occurred in /var/logadm/timestamp like following:
> "/zones/c3f09892-f36e-4784-9853-b4ceb3e762
>
>
> Is there any other way to patch VM.js without the need to reboot the node?
>
>
Hi Stefan,
Yes, there is. Though as usual, I'd usually recommend rebooting as the
easier option wherever possible.
If you do want to mount over VM.js, you can copy the modified file
somewhere such as /tmp/.patche
>
># uname -v
>joyent_20160818T234814Z
>
>
Stefan,
Based on that timestamp it seems one possible explanation for your problems
is https://smartos.org/bugview/OS-5643 which was fixed about one month
after that platform. Do you use KVM VMs? Do you reboot them frequently?
> Also: when this
>
>
> every now and then vmadm operations fail with
>
>zoneevent watcher exited prematurely with code: 1
>
> leaving the caller in limbo about about the state of the vm. Our
> workaround
> is to restart vmadmd with
>
>svcadm restart vmadmd
>
> which terminates not only vmadmd but also all
>
> thanks, works like a charm. Will this patch be included in the upcoming
> 201702016 release?
>
>
Hi Stefan,
Glad to hear that worked for you.
Yes, that change should be included in the 201702016 release.
Thanks for reporting the problem!
Josh
---
s
Stefan,
I have written up a ticket for what I've found at
https://smartos.org/bugview/OS-5950 . I also have a prototype of a fix at:
https://cr.joyent.us/#/c/1483/ which I'll run through some more testing,
but so far does seem to solve the problem you've described.
I don't know how you have thing
>
> I am concerned about why our old recue image did not boot at all and about
> why a considerable amount of time is spent to delay the boot on the other
> images. Nowadays it takes virtually no time to start the operating system
> hence a delay of up to 20 seconds spent in fetching some data fro
Stefan,
As an update to my last email... I spent some time looking at the logs and
the code and I have a theory about what's going on and why the timing has
changed here. I should be able to confirm this tomorrow and will let you
know what I find.
Thanks,
Josh
-
>
>
> you must provide a user_script to set the root password. You may also
> provide root_authorized_keys to enable ssh access.
>
> I did see a POLLERR message, but it's unclear which key it was trying
>> to query at that point. Do you know?
>>
>
> It queries overwrite_root_akeys in set-root-auth
>
>
> Nothing is logged to vmadm console, you must monitor the VNC output
> instead.
>
>
Hi Stefan,
I was able to boot this image, but it appears there's a password required
to login unlike the default RIPLinuX.
I did see a POLLERR message, but it's unclear which key it was trying to
query at tha
> Any ideas what the cause of this strange
> behavior may be?
>
>
The most likely culprit based on the time and description of the problem is
https://smartos.org/bugview/OS-5783 which implemented the changes described
in https://github.com/joyent/rfd/blob/master/rfd/0069/README.md.
In order to det
>
>
> I haven't noticed it until now (probably because I never list all
> snapshots but only snapshots of a single filesystem (-t snapshot -d 1)) but
> /usr/sbin/zfs (ELF 32-bit) runs out of memory.
>
> # /usr/sbin/zfs list -H -p -t filesystem,snapshot,volume -o
> compression,creation,mountpoint,na
>
>
>
> If you look at the output in the last message from the user who closed
> this issue, you'll see that they now have far fewer snapshots than they
> used to. The impression I have there is that this user no longer has the
> problem because they just no longer have enough snapshots for it to b
>
>
> The bug report claims that the problem no longer exists but I'm running
> 20170105T024017Z on the system (2016-06-09 before that, which had the same
> problem.)
>
>
Sven,
If you look at the output in the last message from the user who closed this
issue, you'll see that they now have far fewe
> The system has 901147 snapshots over 106 filesystems and only 16 GB of RAM.
>
> I assume ZFS should easily be able to handle 900'000 snapshots?
>
>
Hi Sven,
The actual command being run is different from what you ran. Can you tell
me what the output is of:
time /usr/sbin/zfs list -H -p -t files
Hello,
A change has just been pushed to master which introduces a flag day between
smartos-live and illumos-joyent. If you don't build your own platforms, or
if you always make sure to update both illumos-joyent and smartos-live
before building, or if you never use LX you shouldn't need to worry a
> Got this when stopping a SmartOS zone, vmadm list returns the same error.
>
> vmadm stop 9652418f-a059-eb30-9024-de07910b0fb0
> Uncaught Error: EIO, i/o error
>
> FROM
> Function.startup.resolveArgv0 (node.js:811:23)
> startup (node.js:58:13)
> node.js:902:3
> Abort (core dumped)
>
This happen
> Thanks for the details. I have some idea what the problem might be but I have
> not had time
> to reproduce this yet.
Alessio,
Sorry it took so long to get back to you.
This should have been fixed in the OS-5442/OS-4897 changes. Please let me know
if you see this issue again on a platform new
Alessio,
Thanks for the details. I have some idea what the problem might be but I have
not had time to reproduce this yet. The procedure you've outlined is not
something that's supported (one should never manually edit/replace the files in
/etc/zones) but I will attempt to confirm the reason it
> Who is responsible of metadata.sock? Which service creates it? Where I
> am wrong?
Hello,
The metadata.sock is created by the svc:/system/smartdc/metadata:default
service running in the GZ.
Could you explain exactly what procedure you ran through to cause this problem?
And what platform vers
> Is there a way to "protect" a SmartOS zone or KVM from being accidently
> deleted?
Hi,
If you search for the indestructible_zoneroot and indestructible_delegated
options in the vmadm man page that might be what you're looking for.
Hope that helps,
Josh
> I'm having issues creating a nginx zone from official image.
[...]
> vmadm create < internal-nginx.json
> timed out waiting for /var/svc/provisioning to move for
> ad0d9a48-d969-e952-c822-939d8ef4f6e9
Hi Roberto,
Can you tell me what's in the following files in that failed zone?
* /var/s
I used vmadm, it works fine for adding but updating/remove is a bit flaky.
Hi Jorge,
Could you be a bit more specific as to what problems you had with vmadm
update/remove?
Thanks,
Josh
---
smartos-discuss
Archives: https://www.listbox.com/member/a
Hi,
I am using VNC to access some Windows vm's and the default keyboard layout is
"en-us" whereas I need to use "en-gb". I can see the "-k en-us" option in the
vm.startvm.log but cannot find the appropriate vmadm json value to change it.
Hi,
If you look at the vmadm man page for qemu_opts tha
>
>
> I found that the VM will reboot once after creation.
> The vm transition is like “provisioning” -> “running” -> “shutting_down”
> -> “stopped” -> “running”
>
> I’d like to know:
>
>1. What happens after provisioning between the first running and the
>second running
>2. Which servi
> I am having problems using vmadm send to a remote host when I have a zone
> configured with the ‘delegate_dataset’ flag. It looks like it doesn’t
> transfer the delegated fs so it ends up throwing errors.
>
> Is this something that will be supported in the future?
>
Hi Anil,
The vmadm send/recv
Hi Eric,
I'm assuming by "reinstall" here you mean that you'd like to just wipe out
and recreate the zone. If that's not what you want, the instructions below
will not be helpful. In that case, could you be a bit more specific?
One option here would be to use sdcadm and 'update' to the same image
>
> Is this a known problem, and is there a fix? I'm not 100% sure how logadm
> is choosing these log files.
>
>
This is a bug in logadm. You can find a few more details in:
https://github.com/joyent/smartos-live/issues/319
It has not been fixed yet but there are some notes in the comments there
> Is there any work being done (if not, I'd be happy to have a look if I
> knew where) to address the painful response times (many minutes in my case)
> for vmadm commands when the system has a large number of snapshots?
>
>
Yes, work is being done but unfortunately progress has been slow due to
ot
>
> ...
> "files": [
>{
> "path": "/tmp/www.image.gz",
> "sha1": "8a9c123d5523f48197c996fd0778e39cd453f6e4",
> "size": "1971920581",
> }
> ],
>
...
> Running imgadm with this manifest and my image results in:
> # imgadm install -m manifest -f image.gz
> imgadm: error (InvalidMa
>
> First time I've used vmadm send/receive, used zfs many times. I don't
> understand what is wrong. Hoping someone can point me the right direction.
>
>
>
...
>
> [root@smartos01 /opt]# vmadm send 91318658-905f-4503-b3ff-180c55e73a7f
> |ssh 10.0.0.5 vmadm receive
> Password:
> Invalid value(s
>
>
> What I'd like to find out is whether there is someplace I can look that
> would help me understand why the shutdowns are taking so long.
>
>
Are these KVM or SmartOS VMs?
If SmartOS, can you try running 'shutdown -y -g 0 -i 5' in one of them that
won't shut down, and tell me whether that wor
>
>
> I always get the message "zoneevent watcher exited prematurely with code:
> 1" when I rollback a local zone or create one.
>
> Commands I use:
>
> vmadm rollback-snapshot
> vmadm create -f *.json
>
> Any one got the same problem or anything helpful known about this message?
>
>
Hi,
One of
> This is a bug. Thanks for pointing this out.
>
> I'll try to get this fixed for the next release. I've filed a ticket
> internally OS-2872.
This should be fixed in master now.
I've also opened up the ticket for bugview as Nahum suggested.
Thanks,
Josh
---
u-dev:KVM
> e12158a6-ddbf-4d1d-b0d8-1eeef8c55f8c:W7x86:KVM
> e7303b27-69b5-437b-b763-f2475b840730:ubuntu:KVM
> edabff7e-5b53-46d2-80aa-9f27b054aa43:centos-dev:KVM
>
> After goring trough git log I found commit after which it stopped working.
>
> commit
> 19463e8765691f980d8944fa91b6d
43 matches
Mail list logo
