[07:18] <Mashkoor> hi team
[07:18] <Mashkoor> can I use MAAS for my LAN Infra for autodeploymnet of different operating systems? like windows and linux
[12:11] <hkominos> Greetings. I am having a weird issue with maas-tftp. It is veeeeeeeeery slow. (The issue does not appear to be networking related. All machines are on the same host). Has anybody seen similar behaviour on aarch64 ?
[12:11] <hkominos> I googled a bit and saw that tftblocksize could be increased to help. Can taht be done on maas-tftp?
[12:56] <roaksoax> hkominos: did you ugprade due to meltdown/spectre ?
[12:56] <roaksoax> hkominos: it could be related to that
[12:56] <roaksoax> hkominos: is your maas running on a VM or on actual hardware?
[13:00] <hkominos> it is running on a VM. and we are booting both physical hardware and Vms (aarch64 all of them). This issue has been here for some time.it takes about 40 seconds to transfer the images from the VM to the machine
[13:00] <hkominos> current kernel is 4.4.0-87
[13:00] <hkominos> ubuntu
[13:40] <roaksoax> kwmonroe: right, but is maas itself, where is it running? on a vm inside a aarch64 ?
[14:30] <dnegreira> Hi, I am having trouble starting MAAS after an upgrade
[14:30] <dnegreira> I am getting this on /var/log/maas/regiond.log: /usr/bin/twistd3: Unknown command: maas-regiond
[14:31] <roaksoax> dnegreira: that seems that your machine didn't really fully upgrade
[14:31] <roaksoax> or something prevented it from fully upgraded
[14:31] <dnegreira> shall I run dpkg --reconfigure ?
[14:31] <dnegreira> on the maas-region-controller package
[14:32] <dnegreira> 2018-01-25 15:22:52 status half-configured maas:all 2.3.0-6434-gd354690-0ubuntu1~16.04.1
[14:32] <dnegreira> you are right indeed
[14:32] <dnegreira> weird, I seemed to miss any message regarding this during the upgrade, I will try to see what happened
[14:35] <roaksoax> dnegreira: try sudo apt-get -f install
[14:37] <dnegreira> reports nothing
[14:38] <dnegreira> also nothing in dpkg -l or dpkg-query
[14:38] <dnegreira> the plot thickens
[14:40] <dnegreira> ok figured it out
[14:40] <dnegreira> it seems that rackd was not able to read out the files in /etc/apt/apt.conf.d/
[14:41] <dnegreira> fixed, and sorry for the noise
[14:45] <roaksoax> nop worries
[15:17] <hkominos> roaksoax: Yes
[15:22] <hkominos> Progress: I have been trying to play around with the blk_size in  is /usr/lib/python3/dist-packages/tftp/bootstrap.py  is 1028 . if i push it a bit further (I assume this leads to bigger ip packets so less time to offload the image) Then tptf stalls and does not serve anything
[15:23] <hkominos> I will see if I can somehow enable jumbo frames
[16:31] <xygnal> roaksoax: mpontillo: i regards to bug 1744765 - it sounds like you confirmed that with over 300 nodes it slowed down a bit? is that something expected? its quite slow for us. need to know if that is being looked into or if we need to split up our environments
[16:40] <roaksoax> xygnal: did you patch your MAAS for metldown ?
[16:40] <roaksoax> err
[16:40] <roaksoax> your system for metldown ?
[16:41] <roaksoax> xygnal: we've tested the listing (both devices and machine listing) with 8k machines wihtout issues
[16:42] <roaksoax> but that was before metldown
[16:42] <roaksoax> xygnal: how many machines do you have ?
[16:42] <roaksoax> xygnal: what's the technical specs of the machine running maas? inside a vm or in hardware ?
[16:42] <xygnal> just under 600 according to my bug report
[16:42] <xygnal> VM
[16:42] <xygnal> ESXi ubuntu vm
[16:42] <xygnal> 8 cores, 32GB of ram
[16:43] <xygnal> storage is NFS-based (As in that is how ESXi is storing its volumes) so we have swap disabled
[16:44] <roaksoax> xygnal: right
[16:44] <roaksoax> xygnal: so did you start seeing this recently? judging by the date of the bug it would seem after metldown fix on the kernel was released ?
[16:45] <xygnal> when NOT refreshing the nodes page, the memory issue doesnt trigger and memory utilization is normal
[16:45] <xygnal> usually under 8gb
[16:45] <xygnal> however
[16:45] <xygnal> it is just as slow then
[16:45] <xygnal> its is slow the whole time
[16:45] <Beret> xygnal, did you file a support request with Canonical?
[16:46] <xygnal> no, we have only worked on this with the MAAS devs so far
[16:47] <xygnal> and as if yet dont feel i have sufficient evidence that that is the reason its performing poorly, unless you have some way to prove thats the behavior its sitting on
[16:48] <Beret> xygnal, quickest way to resolution is to file a support request
[16:49] <xygnal> Beret: we could jsut boot the old kernel on maas 2.3 and see if it peforms any differently
[16:49] <xygnal> far faster than filing a support request to find out
[18:27] <xygnal> boot back for dev environment certainly seems faster, but not ready to test it prod yet
[18:27] <xygnal> just noticed prod is ALSO unable to look up node details
[18:27] <xygnal> if open up individual nodes it just sits at 'loading'
[18:34] <roaksoax> xygnal: yeah improvements for machine details loading are in 2.4
[18:34] <roaksoax> xygnal: but for listing are in 2.3
[18:34] <roaksoax> xygnal: but that's just because it loads the commissioning logs and such
[19:26] <roaksoax> t/win 5