Re: ***UNCHECKED*** Question about 'Deployment Planner'
Hi Haijiao, Try the following global configurations vm.allocation.algorithm = firstfitleastconsumed host.capacityType.to.order.clusters = CPU (or RAM) -Wei Haijiao <18602198...@163.com> 于2019年4月9日周二 上午8:48写道： > > > Even we have ran CloudStack for years,still not quite clear about how > 'First Fit' method exactly works ? > > > It's stated in http://docs.cloudstack.apache.org that if select 'First > Fit' method in 'Deployment Planner', user shall 'places new VMs on the > first host that is found having sufficient capacity to support the VM’s > requirements.' > > > The question is > > > 1. in what order CloudStack checks hosts to find out who is capable > to host the new VM ? > 2. Is there a method ACS can create the new VM on the host which has > less resource utilizatoin, e.g CPU or memory ? That would be particularly > useful to balance the host resource utilization among hosts within same > cluster. > > > Regards, > > > > > >
Re:Re: ***UNCHECKED*** Question about 'Deployment Planner'
Hi, Wei You rock !It's exactly what we want. Would suggest community to update the 'deployment planner' part in the doc, link to these 2 important global parameters. Thanks again , Wei ! 在2019年04月09 15时18分, "Wei ZHOU"写道: Hi Haijiao, Try the following global configurations vm.allocation.algorithm = firstfitleastconsumed host.capacityType.to.order.clusters = CPU (or RAM) -Wei Haijiao <18602198...@163.com> 于2019年4月9日周二 上午8:48写道： > > > Even we have ran CloudStack for years,still not quite clear about how > 'First Fit' method exactly works ? > > > It's stated in http://docs.cloudstack.apache.org that if select 'First > Fit' method in 'Deployment Planner', user shall 'places new VMs on the > first host that is found having sufficient capacity to support the VM’s > requirements.' > > > The question is > > > 1. in what order CloudStack checks hosts to find out who is capable > to host the new VM ? > 2. Is there a method ACS can create the new VM on the host which has > less resource utilizatoin, e.g CPU or memory ? That would be particularly > useful to balance the host resource utilization among hosts within same > cluster. > > > Regards, > > > > > >
Re: Re: ***UNCHECKED*** Question about 'Deployment Planner'
Hi Haijiao, I would suggest you to update the documentation on this particular thing - it's a great way to contribute, considering documentation is always lagging somewhat behind the feature set in CloudStack ? Best Andrija On Tue, 9 Apr 2019 at 10:24, Haijiao <18602198...@163.com> wrote: > Hi, Wei > > > You rock !It's exactly what we want. > > > Would suggest community to update the 'deployment planner' part in the > doc, link to these 2 important global parameters. > > > Thanks again , Wei ! > > > > > > > 在2019年04月09 15时18分, "Wei ZHOU"写道: > > Hi Haijiao, > > Try the following global configurations > > vm.allocation.algorithm = firstfitleastconsumed > host.capacityType.to.order.clusters = CPU (or RAM) > > -Wei > > > > > Haijiao <18602198...@163.com> 于2019年4月9日周二 上午8:48写道： > > > > > > > Even we have ran CloudStack for years,still not quite clear about how > > 'First Fit' method exactly works ? > > > > > > It's stated in http://docs.cloudstack.apache.org that if select 'First > > Fit' method in 'Deployment Planner', user shall 'places new VMs on the > > first host that is found having sufficient capacity to support the VM’s > > requirements.' > > > > > > The question is > > > > > > 1. in what order CloudStack checks hosts to find out who is capable > > to host the new VM ? > > 2. Is there a method ACS can create the new VM on the host which has > > less resource utilizatoin, e.g CPU or memory ? That would be particularly > > useful to balance the host resource utilization among hosts within same > > cluster. > > > > > > Regards, > > > > > > > > > > > > > -- Andrija Panić
Windows on KVM - Windows profile with PV drivers vs Other PV?
Hi, I'm trying to add some Windows templates to my test cloud and by using the "Windows Server 2016" I end up with Intel 1000 NICs instead of virtio ones; this is a bit suboptimal. When I add the template initially I can see I can specify a type of storage - virtio-scsi is what I need, however I can't do the same for the network devices. What does the community recommend here? Any way to force virtio devices on the Windows Server profile or should I go for "Other PV" profiles for Windows guests? -- Sent from the Delta quadrant using Borg technology! Nux! www.nux.ro
Re: Windows on KVM - Windows profile with PV drivers vs Other PV?
Ok, turns out there's a "Windows PV" profile that sorts out my problem. Didn't see that initially. -- Sent from the Delta quadrant using Borg technology! Nux! www.nux.ro - Original Message - > From: "Nux!" > To: "dev" > Cc: "users" > Sent: Tuesday, 9 April, 2019 14:26:40 > Subject: Windows on KVM - Windows profile with PV drivers vs Other PV? > Hi, > > I'm trying to add some Windows templates to my test cloud and by using the > "Windows Server 2016" I end up with Intel 1000 NICs instead of virtio ones; > this is a bit suboptimal. > > When I add the template initially I can see I can specify a type of storage - > virtio-scsi is what I need, however I can't do the same for the network > devices. > > What does the community recommend here? Any way to force virtio devices on the > Windows Server profile or should I go for "Other PV" profiles for Windows > guests? > > -- > Sent from the Delta quadrant using Borg technology! > > Nux! > www.nux.ro
Re: Windows on KVM - Windows profile with PV drivers vs Other PV?
Lucian, Windows PV also enables the KVM HyperV Enlightenment features for Windows guests on 4.12. https://www.lfasiallc.com/wp-content/uploads/2017/11/Use-Hyper-V-Enlightenments-to-Increase-KVM-VM-Performance_Density_Chao-Peng.pdf You can also change the disk controller when creating a template - [cid:1f21ed36-1c9c-401a-bbf2-34bc7b4ce61a] -Si From: Nux! Sent: Tuesday, April 9, 2019 8:53 AM To: dev Cc: users Subject: Re: Windows on KVM - Windows profile with PV drivers vs Other PV? Ok, turns out there's a "Windows PV" profile that sorts out my problem. Didn't see that initially. -- Sent from the Delta quadrant using Borg technology! Nux! www.nux.ro - Original Message - > From: "Nux!" > To: "dev" > Cc: "users" > Sent: Tuesday, 9 April, 2019 14:26:40 > Subject: Windows on KVM - Windows profile with PV drivers vs Other PV? > Hi, > > I'm trying to add some Windows templates to my test cloud and by using the > "Windows Server 2016" I end up with Intel 1000 NICs instead of virtio ones; > this is a bit suboptimal. > > When I add the template initially I can see I can specify a type of storage - > virtio-scsi is what I need, however I can't do the same for the network > devices. > > What does the community recommend here? Any way to force virtio devices on the > Windows Server profile or should I go for "Other PV" profiles for Windows > guests? > > -- > Sent from the Delta quadrant using Borg technology! > > Nux! > www.nux.ro
[GitHub] [cloudstack-documentation] PaulAngus opened a new pull request #39: add L2 to networking overview
PaulAngus opened a new pull request #39: add L2 to networking overview URL: https://github.com/apache/cloudstack-documentation/pull/39 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
Re: Windows on KVM - Windows profile with PV drivers vs Other PV?
What he said ^^^ :) On Tue, Apr 9, 2019, 16:34 Simon Weller wrote: > Lucian, > > Windows PV also enables the KVM HyperV Enlightenment features for Windows > guests on 4.12. > > > https://www.lfasiallc.com/wp-content/uploads/2017/11/Use-Hyper-V-Enlightenments-to-Increase-KVM-VM-Performance_Density_Chao-Peng.pdf > > You can also change the disk controller when creating a template - > > > > -Si > > -- > *From:* Nux! > *Sent:* Tuesday, April 9, 2019 8:53 AM > *To:* dev > *Cc:* users > *Subject:* Re: Windows on KVM - Windows profile with PV drivers vs Other > PV? > > Ok, turns out there's a "Windows PV" profile that sorts out my problem. > Didn't see that initially. > > -- > Sent from the Delta quadrant using Borg technology! > > Nux! > www.nux.ro > > - Original Message - > > From: "Nux!" > > To: "dev" > > Cc: "users" > > Sent: Tuesday, 9 April, 2019 14:26:40 > > Subject: Windows on KVM - Windows profile with PV drivers vs Other PV? > > > Hi, > > > > I'm trying to add some Windows templates to my test cloud and by using > the > > "Windows Server 2016" I end up with Intel 1000 NICs instead of virtio > ones; > > this is a bit suboptimal. > > > > When I add the template initially I can see I can specify a type of > storage - > > virtio-scsi is what I need, however I can't do the same for the network > > devices. > > > > What does the community recommend here? Any way to force virtio devices > on the > > Windows Server profile or should I go for "Other PV" profiles for Windows > > guests? > > > > -- > > Sent from the Delta quadrant using Borg technology! > > > > Nux! > > www.nux.ro >
Re: Loop snapshots are not automatically deleted
Hi Li, both manual and scheduled snapshots work fine on Ceph - CloudStack 4.11.2 and Ceph Mimic 13.2.5 - confirmed today. i.e. [root@ceph1 ~]# rbd snap ls cloudstack/99b3ee82-2b2e-4d7c-85b5-f580adc6f0fb SNAPID NAME SIZE TIMESTAMP 18 fcd4b719-b0e4-4efc-8cd4-04184af19c39 5 GiB Tue Apr 9 11:20:45 2019 19 fbc2280f-cdfd-4897-a134-dc3e0d6319a8 5 GiB Tue Apr 9 12:20:45 2019 20 db742be5-abcb-4f0c-ac51-36150d90f2d5 5 GiB Tue Apr 9 13:20:45 2019 then after few hours [root@ceph1 ~]# rbd snap ls cloudstack/99b3ee82-2b2e-4d7c-85b5-f580adc6f0fb SNAPID NAME SIZE TIMESTAMP 24 76ac671b-948b-4338-be44-3fcf8538f882 5 GiB Tue Apr 9 15:20:45 2019 26 deefaeb4-2cb2-4dae-a786-6d06e1f236e5 5 GiB Tue Apr 9 16:20:46 2019 28 f2997ace-00ee-4d83-bce0-2faee9f41620 5 GiB Tue Apr 9 17:20:46 2019 ^^^ configured to keep last 3 hourly snapshots. You can get into issues deleting old snapshots in cases that some of the previous schedule snapshot jobs didn't complete, or is in ERROR state and similar and then the mess can happen (had this myself some years ago). I would suggest you examine management server logs to understand what goes wrong. Best, Andrija On Tue, 9 Apr 2019 at 02:12, li jerry wrote: > Hi, Andrija > After the snapshot is manually created and deleted, the snapshot > on ceph has been deleted normally. > > > Jerry > > -邮件原件- > 发件人: Andrija Panic > 发送时间: 2019年4月8日 22:12 > 收件人: users > 抄送: dev@cloudstack.apache.org > 主题: Re: Loop snapshots are not automatically deleted > > Hi, > > can you please test with a new volume, make a few snapshots manually, and > then delete one by one, observing if snapshots is deleted from Ceph ? > > Long days ago, this ^^^ was an issue, with serious consequences on > performance - hope it's not back... > > Andrija > > On Mon, 8 Apr 2019 at 14:21, li jerry wrote: > > > Hi,ALL > > > > I created a loop snapshot on cloudstack 4.11.2 +CEPH > > 10.2.11.Execute at 3 am every day, keep 3 copies.After a period of > > observation, I found that the snapshot on cloudstack always shows 3 > > copies.However, ceph found that the snapshot on rbd is added every day > and never deleted. > > > > > > > > -- > > Andrija Panić > -- Andrija Panić
答复: Loop snapshots are not automatically deleted
Hi Andrija Thank you very much for your reply, I will check my log. -邮件原件- 发件人: Andrija Panic 发送时间: 2019年4月10日 0:45 收件人: users 抄送: dev@cloudstack.apache.org 主题: Re: Loop snapshots are not automatically deleted Hi Li, both manual and scheduled snapshots work fine on Ceph - CloudStack 4.11.2 and Ceph Mimic 13.2.5 - confirmed today. i.e. [root@ceph1 ~]# rbd snap ls cloudstack/99b3ee82-2b2e-4d7c-85b5-f580adc6f0fb SNAPID NAME SIZE TIMESTAMP 18 fcd4b719-b0e4-4efc-8cd4-04184af19c39 5 GiB Tue Apr 9 11:20:45 2019 19 fbc2280f-cdfd-4897-a134-dc3e0d6319a8 5 GiB Tue Apr 9 12:20:45 2019 20 db742be5-abcb-4f0c-ac51-36150d90f2d5 5 GiB Tue Apr 9 13:20:45 2019 then after few hours [root@ceph1 ~]# rbd snap ls cloudstack/99b3ee82-2b2e-4d7c-85b5-f580adc6f0fb SNAPID NAME SIZE TIMESTAMP 24 76ac671b-948b-4338-be44-3fcf8538f882 5 GiB Tue Apr 9 15:20:45 2019 26 deefaeb4-2cb2-4dae-a786-6d06e1f236e5 5 GiB Tue Apr 9 16:20:46 2019 28 f2997ace-00ee-4d83-bce0-2faee9f41620 5 GiB Tue Apr 9 17:20:46 2019 ^^^ configured to keep last 3 hourly snapshots. You can get into issues deleting old snapshots in cases that some of the previous schedule snapshot jobs didn't complete, or is in ERROR state and similar and then the mess can happen (had this myself some years ago). I would suggest you examine management server logs to understand what goes wrong. Best, Andrija On Tue, 9 Apr 2019 at 02:12, li jerry wrote: > Hi, Andrija > After the snapshot is manually created and deleted, the > snapshot on ceph has been deleted normally. > > > Jerry > > -邮件原件- > 发件人: Andrija Panic > 发送时间: 2019年4月8日 22:12 > 收件人: users > 抄送: dev@cloudstack.apache.org > 主题: Re: Loop snapshots are not automatically deleted > > Hi, > > can you please test with a new volume, make a few snapshots manually, > and then delete one by one, observing if snapshots is deleted from Ceph ? > > Long days ago, this ^^^ was an issue, with serious consequences on > performance - hope it's not back... > > Andrija > > On Mon, 8 Apr 2019 at 14:21, li jerry wrote: > > > Hi,ALL > > > > I created a loop snapshot on cloudstack 4.11.2 +CEPH > > 10.2.11.Execute at 3 am every day, keep 3 copies.After a period of > > observation, I found that the snapshot on cloudstack always shows 3 > > copies.However, ceph found that the snapshot on rbd is added every > > day > and never deleted. > > > > > > > > -- > > Andrija Panić > -- Andrija Panić
