search for: ksmd

On 07/31/2017 02:55 PM, Michal Hocko wrote: > On Mon 31-07-17 12:13:33, Wei Wang wrote: >> Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and >> shouldn't be given to the host ksmd to scan. > Could you point me where this MADV_DONTNEED is done, please? Sure. It's done in the hypervisor when the balloon pages are received. Please see line 40 at https://github.com/qemu/qemu/blob/master/hw/virtio/virtio-balloon.c > >> Therefore, it is not >> necessary t...

On 07/31/2017 02:55 PM, Michal Hocko wrote: > On Mon 31-07-17 12:13:33, Wei Wang wrote: >> Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and >> shouldn't be given to the host ksmd to scan. > Could you point me where this MADV_DONTNEED is done, please? Sure. It's done in the hypervisor when the balloon pages are received. Please see line 40 at https://github.com/qemu/qemu/blob/master/hw/virtio/virtio-balloon.c > >> Therefore, it is not >> necessary t...

This patch is a revert of 'commit bb01b64cfab7 ("mm/balloon_compaction.c: enqueue zero page to balloon device")' Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and shouldn't be given to the host ksmd to scan. Therefore, it is not necessary to zero ballooned pages, which is very time consuming when the page amount is large. The ongoing fast balloon tests show that the time to balloon 7G pages is increased from ~491ms to 2.8 seconds with __GFP_ZERO added. So, this patch removes the flag. Signed-...

This patch is a revert of 'commit bb01b64cfab7 ("mm/balloon_compaction.c: enqueue zero page to balloon device")' Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and shouldn't be given to the host ksmd to scan. Therefore, it is not necessary to zero ballooned pages, which is very time consuming when the page amount is large. The ongoing fast balloon tests show that the time to balloon 7G pages is increased from ~491ms to 2.8 seconds with __GFP_ZERO added. So, this patch removes the flag. Signed-...

...t; >>>On 07/31/2017 02:55 PM, Michal Hocko wrote: > >>>> >On Mon 31-07-17 12:13:33, Wei Wang wrote: > >>>>> >>Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > >>>>> >>shouldn't be given to the host ksmd to scan. > >>>> >Could you point me where this MADV_DONTNEED is done, please? > >>> > >>>Sure. It's done in the hypervisor when the balloon pages are received. > >>> > >>>Please see line 40 at > >>>https://github.com/...

Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and shouldn't be given to the host ksmd to scan. Therefore, it is not necessary to zero ballooned pages, which is very time consuming when the page amount is large. The ongoing fast balloon tests show that the time to balloon 7G pages is increased from ~491ms to 2.8 seconds with __GFP_ZERO added. So, this patch removes the flag. Signed-...

Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and shouldn't be given to the host ksmd to scan. Therefore, it is not necessary to zero ballooned pages, which is very time consuming when the page amount is large. The ongoing fast balloon tests show that the time to balloon 7G pages is increased from ~491ms to 2.8 seconds with __GFP_ZERO added. So, this patch removes the flag. Signed-...

...hu 03-08-17 19:59:17, Wei Wang wrote: > This patch is a revert of 'commit bb01b64cfab7 ("mm/balloon_compaction.c: > enqueue zero page to balloon device")' > > Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > shouldn't be given to the host ksmd to scan. I find MADV_DONTNEED reference still quite confusing. What do you think about the following wording instead: " Zeroying ballon pages is rather time consuming, especially when a lot of pages are in flight. E.g. 7GB worth of ballooned memory takes 2.8s with __GFP_ZERO while it takes ~4...

...igned to each VM was 4G, 1.5G, 512M, 256M. It was the 256M VM that was started last and that's the one that was shutdown. What are the memory requirements of the host OS? Another thing I've noticed on the node is that as the memory usage increases a lot of processor time is spent running ksmd. Presumably it's spending time reclaiming memory from buffers and cache. Aren't buffers generally used for disk buffering? Does the host OS need to do this given the guest OS will also be buffering the disks? Cheers, Justin. -------------- next part -------------- An HTML attachment was...

On Mon 31-07-17 12:13:33, Wei Wang wrote: > Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > shouldn't be given to the host ksmd to scan. Could you point me where this MADV_DONTNEED is done, please? > Therefore, it is not > necessary to zero ballooned pages, which is very time consuming when > the page amount is large. The ongoing fast balloon tests show that the > time to balloon 7G pages is increased from ~49...

On Mon 31-07-17 15:41:49, Wei Wang wrote: > On 07/31/2017 02:55 PM, Michal Hocko wrote: > >On Mon 31-07-17 12:13:33, Wei Wang wrote: > >>Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > >>shouldn't be given to the host ksmd to scan. > >Could you point me where this MADV_DONTNEED is done, please? > > Sure. It's done in the hypervisor when the balloon pages are received. > > Please see line 40 at > https://github.com/qemu/qemu/blob/master/hw/virtio/virtio-balloon.c Thanks. Are all hypervisors...

On 03.08.2017 13:59, Wei Wang wrote: > This patch is a revert of 'commit bb01b64cfab7 ("mm/balloon_compaction.c: > enqueue zero page to balloon device")' > > Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > shouldn't be given to the host ksmd to scan. Therefore, it is not > necessary to zero ballooned pages, which is very time consuming when > the page amount is large. The ongoing fast balloon tests show that the > time to balloon 7G pages is increased from ~491ms to 2.8 seconds with > __GFP_ZERO added. So, this patch remove...

...Wang wrote: > > This patch is a revert of 'commit bb01b64cfab7 ("mm/balloon_compaction.c: > > enqueue zero page to balloon device")' > > > > Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > > shouldn't be given to the host ksmd to scan. Therefore, it is not > > necessary to zero ballooned pages, which is very time consuming when > > the page amount is large. The ongoing fast balloon tests show that the > > time to balloon 7G pages is increased from ~491ms to 2.8 seconds with > > __GFP_ZERO added. S...

...Wang wrote: > > This patch is a revert of 'commit bb01b64cfab7 ("mm/balloon_compaction.c: > > enqueue zero page to balloon device")' > > > > Ballooned pages will be marked as MADV_DONTNEED by the hypervisor and > > shouldn't be given to the host ksmd to scan. Therefore, it is not > > necessary to zero ballooned pages, which is very time consuming when > > the page amount is large. The ongoing fast balloon tests show that the > > time to balloon 7G pages is increased from ~491ms to 2.8 seconds with > > __GFP_ZERO added. S...

...agate SATA drive, non-raid. > > > > I have this server which has been dying randomly, with no logs. > > Here is a console picture. > > http://i.imgur.com/ZYHlB82.jpg Thanks to netconsole, I have the panic to post: Feb 16 06:06:56 BUG: soft lockup - CPU#0 stuck for 67s! [ksmd:88] Feb 16 06:06:56 Modules linked in: Feb 16 06:06:56 nf_nat Feb 16 06:06:56 mpt3sas Feb 16 06:06:56 mpt2sas Feb 16 06:06:56 raid_class Feb 16 06:06:56 mptctl Feb 16 06:06:56 ipmi_si Feb 16 06:06:56 ipmi_devintf Feb 16 06:06:56 netconsole Feb 16 06:06:56 configfs Feb 16 06:06:56 ebtable_...

NOTE: this is happening on Centos 6 x86_64, 2.6.32-504.3.3.el6.x86_64 not Centos 5 Dell PowerEdge 2970, Seagate SATA drive, non-raid. I have this server which has been dying randomly, with no logs. I had a tail -f over ssh for a week, when this just happened. Feb 8 00:10:21 thirteen-230 kernel: mptscsih: ioc0: attempting task abort! (sc=ffff880057a0a080) Feb 8 00:10:21 thirteen-230 kernel:

...misc/20 254 ? 00:00:00 md_misc/21 255 ? 00:00:00 md_misc/22 256 ? 00:00:00 md_misc/23 257 ? 00:00:00 khungtaskd 258 ? 00:00:00 kswapd0 259 ? 00:00:00 kswapd1 260 ? 00:00:00 kswapd2 261 ? 00:00:00 kswapd3 262 ? 00:00:00 ksmd 263 ? 00:00:00 khugepaged 264 ? 00:00:00 aio/0 265 ? 00:00:00 aio/1 266 ? 00:00:00 aio/2 267 ? 00:00:00 aio/3 268 ? 00:00:00 aio/4 269 ? 00:00:00 aio/5 270 ? 00:00:00 aio/6 271 ? 00:00:00 aio/7 272 ? 00:00:00...

...0.0 0.0 0 0 ? S 01:38 0:00 \_ [kondemand/3] root 48 0.0 0.0 0 0 ? S 01:38 0:00 \_ [khungtaskd] root 49 0.0 0.0 0 0 ? S 01:38 0:02 \_ [kswapd0] root 50 0.0 0.0 0 0 ? SN 01:38 0:00 \_ [ksmd] root 51 0.0 0.0 0 0 ? S 01:38 0:00 \_ [aio/0] root 52 0.0 0.0 0 0 ? S 01:38 0:00 \_ [aio/1] root 53 0.0 0.0 0 0 ? S 01:38 0:00 \_ [aio/2] root 54 0.0 0.0 0 0 ? S 01:38 0:00 \...

Hello Community, I encountered an error while running "xm console domU". Here is the running script: # xm console domU xenconsole: Could not read tty from store: No such file or directory I googled and found out that this issue is mostly result from forgetting starting xenconsole service. However, after I started xenconsole, the issue is still there. Please see: #