Hi folks I would to know if somebody here use remus with xen-unstable... I deploy two servers with this tools and see some abnormal behavior. I have this xen HVM: kernel = "/usr/lib/xen/boot/hvmloader" device_model = ''/usr/lib/xen/bin/qemu-dm'' builder=''hvm'' memory = 512 shadow_memory = 16 name = "Windows-2003" vif=[''mac=00:13:3e:37:9f:9e,bridge=xenbr0''] disk=[''phy:/dev/drbd0,hda,w'', ''file:/storage/ISOS/win2003-STR2-i386.iso,hdc:cdrom,r''] boot="c" on_poweroff = ''destroy'' on_reboot = ''restart'' n_crash = ''restart'' sdl=0 vnc=1 vnclisten="0.0.0.0" stdvga=0 serial=''pty'' I success create a vm on server xen-0 (my master node). I success migrate the vm to xen-1 (my slave node). I came back successfully vm to xen-0. When I run remus -i 100 Windows-2003 xen-slave, I see the vm on slave node turn paused and vm started on master node. I see remus continuously send packages to slave node. But after 3 or 5 minutes, I see the the VM on master node enter in suspend mode without any human intervention! :) I see that state of vm on slave node has change also, to started mode ( xm list show a ---b-- state). After this point, the vm freeze and everything goes to down!!!! Some details about my enviroment: Two Dell PowerEdge T300 with 1 GB of memory (I know! It''s not enoght) and 160 GB of hard disk. CentOS x86_64 with kernel 2.6.18.8 compile against xen-unstable... I have a isolated network to performer live migration and remus... I hope to be clear here.... Thanks for any help -- *Gilberto Nunes Ferreira* *TI* *Selbetti Gestão de Documentos* *Telefone: +55 (47) 3441-6004* *Celular: +55 (47) 8861-6672* <>< _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
HI again I recompile xen-unstable, and at this time, I have remus running with flag --no-net and everything seems good... I just see that VM on master node change your state between -b----, r----- and ---s--, in feel seconds... I do not know if this is a desired behavior or not... Thanks a lot gilberto nunes escreveu:> Hi folks > > > I would to know if somebody here use remus with xen-unstable... > > I deploy two servers with this tools and see some abnormal behavior. > > I have this xen HVM: > > kernel = "/usr/lib/xen/boot/hvmloader" > device_model = ''/usr/lib/xen/bin/qemu-dm'' > builder=''hvm'' > memory = 512 > shadow_memory = 16 > name = "Windows-2003" > vif=[''mac=00:13:3e:37:9f:9e,bridge=xenbr0''] > disk=[''phy:/dev/drbd0,hda,w'', > ''file:/storage/ISOS/win2003-STR2-i386.iso,hdc:cdrom,r''] > boot="c" > on_poweroff = ''destroy'' > on_reboot = ''restart'' > n_crash = ''restart'' > sdl=0 > vnc=1 > vnclisten="0.0.0.0" > stdvga=0 > serial=''pty'' > > I success create a vm on server xen-0 (my master node). > I success migrate the vm to xen-1 (my slave node). > I came back successfully vm to xen-0. > > When I run remus -i 100 Windows-2003 xen-slave, I see the vm on slave > node turn paused and vm started on master node. > I see remus continuously send packages to slave node. > But after 3 or 5 minutes, I see the the VM on master node enter in > suspend mode without any human intervention! :) > I see that state of vm on slave node has change also, to started mode > ( xm list show a ---b-- state). > After this point, the vm freeze and everything goes to down!!!! > > Some details about my enviroment: > > Two Dell PowerEdge T300 with 1 GB of memory (I know! It''s not enoght) > and 160 GB of hard disk. > CentOS x86_64 with kernel 2.6.18.8 compile against xen-unstable... > I have a isolated network to performer live migration and remus... > > I hope to be clear here.... > > Thanks for any help > > > > > > > >-- *Gilberto Nunes Ferreira* *TI* *Selbetti Gestão de Documentos* *Telefone: +55 (47) 3441-6004* *Celular: +55 (47) 8861-6672* <>< _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
gilberto nunes
2009-Dec-18 13:43 UTC
Re: [Xen-devel] Xen-unstable and Remus (It''s drive me crazy...)
Hi again... Now, I active -i 1000 on remus tool and I have VM on both sides with -b---- state! This is a normal behavior??? thanks gilberto nunes escreveu:> HI again > > I recompile xen-unstable, and at this time, I have remus running with > flag --no-net and everything seems good... > > I just see that VM on master node change your state between -b----, > r----- and ---s--, in feel seconds... > > I do not know if this is a desired behavior or not... > > > Thanks a lot > > gilberto nunes escreveu: >> Hi folks >> >> >> I would to know if somebody here use remus with xen-unstable... >> >> I deploy two servers with this tools and see some abnormal behavior. >> >> I have this xen HVM: >> >> kernel = "/usr/lib/xen/boot/hvmloader" >> device_model = ''/usr/lib/xen/bin/qemu-dm'' >> builder=''hvm'' >> memory = 512 >> shadow_memory = 16 >> name = "Windows-2003" >> vif=[''mac=00:13:3e:37:9f:9e,bridge=xenbr0''] >> disk=[''phy:/dev/drbd0,hda,w'', >> ''file:/storage/ISOS/win2003-STR2-i386.iso,hdc:cdrom,r''] >> boot="c" >> on_poweroff = ''destroy'' >> on_reboot = ''restart'' >> n_crash = ''restart'' >> sdl=0 >> vnc=1 >> vnclisten="0.0.0.0" >> stdvga=0 >> serial=''pty'' >> >> I success create a vm on server xen-0 (my master node). >> I success migrate the vm to xen-1 (my slave node). >> I came back successfully vm to xen-0. >> >> When I run remus -i 100 Windows-2003 xen-slave, I see the vm on slave >> node turn paused and vm started on master node. >> I see remus continuously send packages to slave node. >> But after 3 or 5 minutes, I see the the VM on master node enter in >> suspend mode without any human intervention! :) >> I see that state of vm on slave node has change also, to started mode >> ( xm list show a ---b-- state). >> After this point, the vm freeze and everything goes to down!!!! >> >> Some details about my enviroment: >> >> Two Dell PowerEdge T300 with 1 GB of memory (I know! It''s not enoght) >> and 160 GB of hard disk. >> CentOS x86_64 with kernel 2.6.18.8 compile against xen-unstable... >> I have a isolated network to performer live migration and remus... >> >> I hope to be clear here.... >> >> Thanks for any help >> >> >> >> >> >> >> >> >-- *Gilberto Nunes Ferreira* *TI* *Selbetti Gestão de Documentos* *Telefone: +55 (47) 3441-6004* *Celular: +55 (47) 8861-6672* <>< _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On Friday, 18 December 2009 at 11:29, gilberto nunes wrote:> HI again > > I recompile xen-unstable, and at this time, I have remus running with > flag --no-net and everything seems good... > > I just see that VM on master node change your state between -b----, > r----- and ---s--, in feel seconds... > > I do not know if this is a desired behavior or not...Yes, that''s normal. It should go between r (or b) and s at the interval you''ve specified with -i (200ms, or 5/sec by default). _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Brendan Cully
2009-Dec-18 20:46 UTC
Re: [Xen-devel] Xen-unstable and Remus (It''s drive me crazy...)
On Friday, 18 December 2009 at 11:43, gilberto nunes wrote:> Hi again... > > Now, I active -i 1000 on remus tool and I have VM on both sides with > -b---- state! > > This is a normal behavior???No, that happens if the backup thinks that the primary failed, but it hasn''t. In this case, it''s because the primitive heartbeat built into Remus expects traffic at least every 500ms, so at 1000 it''s not getting it. I''d recommend keeping -i at or below the default of 200 until we make the heartbeat smarter. You could also use something like linux-HA''s STONETH to kill off the primary before unleashing the backup on the world. Seamless integration with linux-ha is a TODO item. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
gilberto nunes
2009-Dec-18 20:52 UTC
Re: [Xen-devel] Xen-unstable and Remus (It''s drive me crazy...)
Thanks for answers I get success work withou -i option... If I put -i flags, with 200 or minor, whatever, remus do not work properly.... Do you have any idea when remus go up to production state? Any date?? Thanks again Brendan Cully escreveu:> On Friday, 18 December 2009 at 11:43, gilberto nunes wrote: > >> Hi again... >> >> Now, I active -i 1000 on remus tool and I have VM on both sides with >> -b---- state! >> >> This is a normal behavior??? >> > > No, that happens if the backup thinks that the primary failed, but it > hasn''t. In this case, it''s because the primitive heartbeat built into > Remus expects traffic at least every 500ms, so at 1000 it''s not > getting it. I''d recommend keeping -i at or below the default of 200 > until we make the heartbeat smarter. You could also use something like > linux-HA''s STONETH to kill off the primary before unleashing the > backup on the world. Seamless integration with linux-ha is a TODO > item. > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel >-- *Gilberto Nunes Ferreira* *TI* *Selbetti Gestão de Documentos* *Telefone: +55 (47) 3441-6004* *Celular: +55 (47) 8861-6672* <>< _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
gilberto nunes
2009-Dec-18 21:10 UTC
Re: [Xen-devel] Xen-unstable and Remus (It''s drive me crazy...)
Hi Brendan
I decrease -i value to 150 and now the state of VM on master is r-----
or ------, and on other side is --b---, but when I destroy VM running
on master, the VM on slave turns to ------ and die! Crazy! rsrs... Any
idea??? Perhaps I do something wrong...
Other doubt: what it means of --port=PORT flag?
Brendan Cully escreveu:
On Friday, 18 December 2009 at 11:43, gilberto nunes wrote:
Hi again...
Now, I active -i 1000 on remus tool and I have VM on both sides with
-b---- state!
This is a normal behavior???
No, that happens if the backup thinks that the primary failed, but it
hasn''t. In this case, it''s because the primitive heartbeat
built into
Remus expects traffic at least every 500ms, so at 1000 it''s not
getting it. I''d recommend keeping -i at or below the default of 200
until we make the heartbeat smarter. You could also use something like
linux-HA''s STONETH to kill off the primary before unleashing the
backup on the world. Seamless integration with linux-ha is a TODO
item.
--
Gilberto Nunes Ferreira
TI
Selbetti Gestão de Documentos
Telefone: +55 (47) 3441-6004
Celular: +55 (47) 8861-6672
<><
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel