Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • May 26 12:04
    marcheschi commented #583
  • May 26 11:49
    marcheschi commented #583
  • May 26 11:44
    marcheschi edited #583
  • May 26 11:39
    marcheschi opened #583
  • May 09 07:30
    marcheschi closed #579
  • May 09 07:30
    marcheschi commented #579
  • May 08 14:24
    YanChii commented #579
  • May 06 12:04
    marcheschi commented #579
  • May 05 13:50
    marcheschi commented #579
  • May 05 13:22
    marcheschi commented #579
  • May 05 13:11
    marcheschi commented #579
  • May 05 13:02
    marcheschi closed #582
  • May 05 13:00
    marcheschi opened #582
  • Apr 19 17:04
    YanChii commented #581
  • Apr 19 17:03
    YanChii labeled #581
  • Apr 19 17:03
    YanChii labeled #581
  • Apr 19 17:03
    YanChii opened #581
  • Apr 13 09:03
    marcheschi opened #580
  • Apr 05 08:39
    marcheschi commented #579
  • Apr 05 08:22
    YanChii commented #579
Jan Poctavek
@YanChii
https://www.reddit.com/r/zfs/comments/opu43n/zvol_used_size_far_greater_than_volsize/
Paolo Marcheschi
@marcheschi
Hi @YanChii , we'd like to add a volume as sparse volume with no reservation in order to move data from one volume to the other , is there a way to attach to the VM on Danubecloud ?
yanchii
@yanchii:matrix.org
[m]

Hi, Paolo.
You can use api with refreservation defined
https://docs.danubecloud.org/api-reference/api/vm_define.html#vm-hostname-or-uuid-define-disk-disk-id

Or just do zfs set refreservation=None <fs> after deploy. You can create smaller disk and grow it later.

by api, I mean es command
Paolo Marcheschi
@marcheschi
I tried with es './es create /vm/e7f3d7d0-d2ca-4332-98g8-7t1bd4eb2432/define/disk/4 -size 52428800 -boot false -block_size 131072 -refreservation 0'
but it says size": [
"Not enough free disk space on node."
]
while with zfs I can create a sparse volume with 90 TB size
yanchii
@yanchii:matrix.org
[m]
change temporarily the storage coefficient
Node->Storage->Edit->Advanced
Alessio C
@alessioftgm_gitlab
and there is no way to "attach" an existing disk to an existing VM?
yanchii
@yanchii:matrix.org
[m]
not directly… you can create a new disk with the same size and either clone the snapshot from source VM to the new disk, or just restore it from backup
if you want to be more adventurous, create new vm on the same CN, zfs delete it's disk and zfs rename the source disk to the name of the deleted volume
wilbury
@wilbury:matrix.org
[m]
will erigonesd pick up that change automagically?
Paolo Marcheschi
@marcheschi

Node->Storage->Edit->Advanced

not possible: Value is too low because of existing virtual machines.

Paolo Marcheschi
@marcheschi

will erigonesd pick up that change automagically?

no I tried

yanchii
@yanchii:matrix.org
[m]
Paolo: increase it much more. See the available storage number in Node's Storages tab
wilbury: erigonesd/vmadm don't care as long as the zvol names don't change. If you however change the zvol size, you might need to re-read the VM state from node (there's an api command for that)
Paolo Marcheschi
@marcheschi
@yanchii:matrix.org you are right I created another volume of 90 TB
wilbury
@wilbury:matrix.org
[m]
what could PROBLEM: Queue mgmt is not responding to ping mean?
Paolo Marcheschi
@marcheschi
@wilbury:matrix.org the node (erigonesd on a node) connects to the mgmt01 and responds to requests so if node is not reachable, I expect some failed erigonesd on the node
you can check services: svcs erigonesd
yanchii
@yanchii:matrix.org
[m]
wilbury: check erigonesd in the mgmt01 VM.
Plus to check the MQ itself, run on a node:
erigonesd inspect ping
you should be able to see all erigonesd daemons in the reply, including the mgmt
balwo
@balwo

Hi,

I'm trying to install danube cloud 4.6 compute-node (no 4.6 head-node available ?) on a Gigabyte BRIX GB-BRR7-4800.
Detailed specs are available at: https://www.gigabyte.com/be/Mini-PcBarebone/GB-BRR7-4800-rev-10#ov

It looks like I'm running into an issue with the onboard NIC during the installation.

When I boot the pc, I get to the Danube boot menu and booting starts.

Loading unix...
Loading /platform/i86pc/amd64/boot_archive...
Loading /platform/i86pc/amd64/boot_archive.hash...
Booting...
WARNING: Couldn't read ACPI SRAT table from BIOS. lgrp... -> this is no issue ; I see the same on my NUC

Illuminos Version esdc_20220127T211339Z 64-bit
ERROR: failed to attach AMD IOMMU -> Seems not to be an issue (https://illumos.topicbox.com/groups/developer/Ted97385c7e5bfb1f/illumos-8679-failed-to-attach-amd-iommu)
Running from USB...
skipping USB copy setup: no zones zpool
Error: unable to initialize empty mac!
/usr/bin/nictagadm: line 212: LINKS[$mac]: bad array subscript
ERROR: cannot configure the system, no NICs were found.
May 11... svc.startd[8]: svc:/network/physical:default: Method "/lib/svc/method/net-physical" failed with exit status 96
May 11... svc.startd[8]: network/physical:default misconfigured: transitioned to maintenance. (see 'svcs -xv' ...)

Some more services (missing hostname, audit,...) report an issue, and I end at the prompt:

Welcome to Danube Cloud node (20220127T211339Z)
unknown wscons login:

Any idea how to get around this ?

Paolo Marcheschi
@marcheschi
Hi @balwo more than a Danubecloud issue is a Smartos issue, it seems that this kind of minipc is not supported by illumos, see the illumos Hardware Compatibility List https://illumos.org/hcl/
Paolo Marcheschi
@marcheschi
Hi I'd like to create a shared NFS from the global zone of a node to be used in a bhyve VM of the same node, is there a way to be sure not to pass from the external network to do data transfers ?
Jan Poctavek
@YanChii
you can temporarily configure new node NIC to be in the same subnet as the VM
dladm create-vnic -l uplink myvlan123 && ipadm create-if myvlan123
ipadm create-addr -T static -a 1.2.3.4/24 myvlan123/sometag
"uplink" is NIC name from dladm show-phys
and create-vnic supports vlan number, e.g. -v 123 for vlan 123
Paolo Marcheschi
@marcheschi
thank you @YanChii this is very interesting, but very difficult for me , I have to practice first ;)
yanchii
@yanchii:matrix.org
[m]
alternatively, add admin vnic to the VM :)
balwo
@balwo
Thank you, Paolo. The realtek network chipset is apparently not supported. That's a pity.
Paolo Marcheschi
@marcheschi

alternatively, add admin vnic to the VM :)

Thank you again @YanChii

@YanChii I have a backup task that is taking log , is there a possibility to see what is doing ?
Paolo Marcheschi
@marcheschi
I found it myself
it is enough to see the backup file that is growing
yanchii
@yanchii:matrix.org
[m]
yup, that is a good way to check. And you can see also esbackup process on the destination server.
Paolo Marcheschi
@marcheschi

we have found a NFS performance issue connected with bhyve, with NFS, the Vm is stuck and it does not work
https://smartos.topicbox.com/groups/smartos-discuss/Ta60043d861986cca

It is caused by the
NAPI tx mode in virtio_net, which seems to conflict with bhyve somehow.
As a workaround, it can be disabled again by running:

ethtool -C net0 tx-frames 0

inside the bhyve VM

Paolo Marcheschi
@marcheschi
In order to increase transfer speed Dan McDonald suggested me to issue these commands;
ipadm set-prop -p max_buf=8388608 tcp
ipadm set-prop -p recv_buf=2097152 tcp
ipadm set-prop -p send_buf=2097152 tcp
these settings does not persist on reboot, is there a way to put them in the usbkey?
wilbury
@wilbury:matrix.org
[m]
there is a recommended way to create custom smf manifest
wilbury
@wilbury:matrix.org
[m]
and put it into /opt/custom/smf so it gets loaded upon boot
Paolo Marcheschi
@marcheschi
Thank you I will try @wilbury:matrix.org
Jan Poctavek
@YanChii
we already have the custom RC scripts in DC. Place the script into /opt/custom/etc/rc-pre-vmadmd.d/ (see example scripts that are already there).
The order of execution is
rc-pre-network.d -> networking -> rc-pre-vmadmd.d -> start VMs -> rc-post-vmadmd.d
Paolo Marcheschi
@marcheschi
This is great @YanChii DC is full of resources
Paolo Marcheschi
@marcheschi
Hi @YanChii I find out that the '/opt/custom/etc/rc-pre-vmadmd.d/' folder is not present on converted smartos systems, is there a script to build the necessary folder and files on the node?