00:53:22 <tozhu> hello experts, I want’ to change zfs_arc_max  value in smartos without reboot the machine, may I know what’s the command used to do that, and how to verify it is changed successed ?
00:53:29 <tozhu> thanks
01:16:17 <tozhu> and also, would like to ask a question, regarding to sys usage is too high when there is more than 10 process running, here is the screen dump by vmstat;  https://pastebin.com/hewQ4WPK, how to avoid it the high sys usage? how to do more investigative in this case? thanks in advance 
02:56:10 <jclulow> tozhu: I would do some kernel stack profiling during that window to see what the kernel is doing
02:56:34 <jclulow> e.g., see https://github.com/brendangregg/FlameGraph#dtrace
02:59:18 <tozhu> jclulow: Thanks for the advice, btw, may I ask how to change ‘zfs_arc_max’  by mdb command without reboot the OS? I read some docs say: echo ‘::zfs_arc_max/W0xxxxx’ | mdb -kw     ; but I don’t know /W0xxxx means,  where is the docs for the details ?
02:59:19 <tozhu> thank you
02:59:52 <jclulow> I do not believe you can change that value at runtime
03:00:19 <jclulow> Or, more accurately, you can probably change it, but it won't do anything
03:00:36 <jclulow> I'm pretty sure `zfs_arc_max` is used once at boot in `arc_init()` 
03:00:47 <jclulow> You will need to change it in `/etc/system`
03:13:44 <tozhu> jclulow: two question for advice: 1.) my server configuration: CPU: 2 socket/packages,  16c/32t per socket,     32c/64t in total;   memory is 384G;  storage:  24x1.2T at mirror ;   this is my server configuration, my question is,  how to allowcate vCPU and Memory for a VM,   do you have any advice?   I use 1vCPU manage 4G mem, usually, a VM had 4vCPU and 16G mem   ;     2.) how many memory is adviced for zfs_arc_max
03:13:53 <tozhu> than you; 
03:14:49 <tozhu> btw,  ‘/etc/system’ is read only for smartos, is there any entry to change it for perim usage?   
03:18:05 <tozhu>  may I ask what’s the new entry for this old one? rsync://pkgsrc.smartos.org/pkgsrc-joyent   do you know?  thanks again
03:19:09 <tozhu> sorry, the old name is pkgsrc.joyent.com , may I know what’s the new entry to do rsync, thank you
03:23:27 <jclulow> tozhu: Ideally you would not have to tune zfs_arc_max
03:23:59 <jclulow> Are you experiencing an issue that makes you feel like you need to tune it?
03:26:13 <tozhu> yes, the situation is that: I’m stress a postgresql database, the database is used about 48G memory, and the zfs_arc_max used 280G memory by arcstat command, I don’t know why zfs_arc_max eat so much memory
03:29:51 <jclulow> It will tend to grow to use unused memory
03:30:14 <jclulow> If the database asks for more memory it should begin to shrink
03:30:37 <tozhu> okay, got it, thank you very for the advice
03:32:05 <jclulow> If it doesn't shrink though, we should investigate why not
03:37:30 <tozhu> jclulow, thank you very much
03:46:06 <jclulow> You're welcome!
06:44:36 <jperkin> tozhu: it's now just rsync://pkgsrc.smartos.org/pkgsrc - note that if you only provide a URL to rsync you can use it like a file browser, e.g. running "rsync rsync://pkgsrc.smartos.org/" will show you the list of available repositories, and then "rsync rsync://pkgsrc.smartos.org/pkgsrc" will list that repo
06:46:07 <tozhu> jperkin: thank you, does this works for the old version of smartos images/zones ?
06:46:36 <jperkin> could you clarify what you mean?
06:49:40 <tozhu> jperkin: I mean if this site  is the full mirror for pkgsrc.joyent.com ?
07:01:43 <jperkin> mostly, yes, if there are old releases that are no longer available then you'll need to grab them from the archive (using manta-sync) at https://us-central.manta.mnx.io/pkgsrc/public/packages/
15:24:54 <blackwood821> Hi, how can I forced delete a zone? CloudAPI failed, VMAPI failed and even `vmadm delete` is hanging...
15:25:35 <bahamat> Is it running?
15:25:47 <blackwood821> Can I destroy the ZFS datasets and remove the /etc/zones/<uuid>.xml file?
15:26:04 <blackwood821> It is running but it has no NICs. It looks like it was never properly created
15:26:25 <blackwood821> There are some old jobs that say the zone was successfully destroyed but it still exists and is running with no NICs
15:26:37 <bahamat> Use vmadm stop -F first
15:27:01 <bahamat> Once it's stopped, there shouldn't be any problem deleting it.
15:27:26 <blackwood821> I get "cannot manage a zone which is in state 'configured'" when I run `vmadm stop -F` but `vmadm list` shows it's in the "running" state
15:27:40 <blackwood821> Should I manually change the state in /etc/zones/<uuid>.xml?
15:28:59 <blackwood821> Actually, I don't see any state/status/transition in the zone XML file
15:45:16 <blackwood821> `zoneadm -z <vm_uuid> halt` worked followed by `vmadm delete <vm_uuid>`