Re: [pve-devel] savevm tests

2012-08-29 Thread Dietmar Maurer
I guess we will always restart the VM with correct configuration when we do a rollback? Oh yes, we need to remove the disk from the conf. But not delete it if we want to rollback again to snap2. yes (as already suggested, we save the whole configuration when creating a snapshot)

Re: [pve-devel] savevm tests

2012-08-29 Thread Dietmar Maurer
* create_snapshot: (savevm?) Yes, if storage support it (qcow2,rbd,sheepdog). I would like to implement also snasphot for external storage like nexenta We simply need to define a create_snapshot sub for each storage module But we need that inside kvm?

Re: [pve-devel] savevm tests

2012-08-29 Thread Dietmar Maurer
* create_snapshot: (savevm?) Yes, if storage support it (qcow2,rbd,sheepdog). I would like to implement also snasphot for external storage like nexenta We simply need to define a create_snapshot sub for each storage module I am quite unsure about that, So far, we have many storage types which

Re: [pve-devel] savevm tests

2012-08-29 Thread Dietmar Maurer
Putting qcow2 (or better qcow3) on those devices would solve the problem? If that works, there is no more need to use nexenta snapshot support? I want to use iscsi with nexenta, so qcow2 is not an option. (and I'm not sure using qcow2 on block device is possible) I see no real reason why

Re: [pve-devel] ahci unmigratable

2012-08-29 Thread Dietmar Maurer
Well, after review I do not like all that code anymore ;-) what if we add an 'ahci' option instead: ahci: 0|1 If set, we use ahci/sata for ide drives, else we use normal ide mode. That way we have the same logic as we use for scsi (scsihw). We would also save some pci addresses. Using ide

Re: [pve-devel] [PATCH] - preserve authorized_key key order - identify double keys by key and not by comment

2012-08-29 Thread Dietmar Maurer
Subject: [pve-devel] [PATCH] - preserve authorized_key key order - identify double keys by key and not by comment Signed-off-by: Stefan Priebe s.pri...@profihost.ag --- data/PVE/Cluster.pm | 26 -- 1 file changed, 12 insertions(+), 14 deletions(-) diff --git

Re: [pve-devel] [PATCH] - preserve authorized_key key order - identify double keys by key and not by comment

2012-08-29 Thread Dietmar Maurer
Ok, committed - please review and test: https://git.proxmox.com/?p=pve-cluster.git;a=commitdiff;h=2055b0a9e41912cb02810b621608b24430c8a1fe -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Dietmar Maurer Sent

[pve-devel] live migration problems

2012-08-29 Thread Dietmar Maurer
The bug is hard to reproduce . We still not found a reliable way to reproduce it. https://bugzilla.proxmox.com/show_bug.cgi?id=247 Any ideas? - Dietmar ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] live migration problems

2012-08-30 Thread Dietmar Maurer
so you can reproduce the bug ? One idea, with my patch,the vm config file is on source during migration and stats are done on disk with qmp during live migration. Maybe this can cause some queues in multiplexing part ? Maybe, but my recent patches should avoid that problem? I disabled

Re: [pve-devel] live migration problems

2012-08-30 Thread Dietmar Maurer
id 103 -chardev socket,id=qmp,path=/var/run/qemu- server/103.qmp,server,nowait -mon chardev=qmp,mode=control - Maybe we should use mux=on? Tried that, but monitor is still limited to one connection. Is that intentional, or a bug? ___ pve-devel

Re: [pve-devel] [PATCH] added renderer for hd strings in hardware overview

2012-08-30 Thread Dietmar Maurer
Oh sorry. I'm travelling right now and Timo told me it looks like: 1024.000 Maybe the automatic update from bps to mbps can cause such numbers. for 1024MB/s. But normally, you simply get what you type on the GUI. ___ pve-devel mailing list

Re: [pve-devel] Fwd: [Qemu-devel] [PATCH] ahci: add migration support

2012-08-31 Thread Dietmar Maurer
Ok, will try to add that -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre DERUMIER Sent: Freitag, 31. August 2012 10:35 To: pve-devel@pve.proxmox.com Subject: [pve-devel] Fwd: [Qemu-devel] [PATCH] ahci: add

Re: [pve-devel] savevm tests

2012-08-31 Thread Dietmar Maurer
I thing I didn't have check, is where the vmstate is stored if we have multiple devices. On the first disk ? on the boot disk ? I think it is the first drive able to create snapshots. They simply do a stupid loop in block.c: BlockDriverState *bdrv_snapshots(void) { BlockDriverState

Re: [pve-devel] Fwd: [Qemu-devel] [PATCH] ahci: add migration support

2012-08-31 Thread Dietmar Maurer
oh sorry - fixed -Original Message- From: Alexandre DERUMIER [mailto:aderum...@odiso.com] Sent: Freitag, 31. August 2012 11:23 To: Dietmar Maurer Cc: pve-devel@pve.proxmox.com Subject: Re: [pve-devel] Fwd: [Qemu-devel] [PATCH] ahci: add migration support don't build

[pve-devel] create snapshots

2012-08-31 Thread Dietmar Maurer
I just looked at savevm, and that is relatively simple code: qemu-kvm/savevm.c: do_savevm() 1.)check if all disk support snapshots (call bdrv_can_snapshot()) 2.)vm_stop(RUN_STATE_SAVE_VM) 3.)Delete old snapshots of the same name 4.)save the VM state 5.)create the

Re: [pve-devel] create snapshots

2012-08-31 Thread Dietmar Maurer
I wonder it is would be safe to call an external script to create the snapshot there? Or is there still some data in the qemu block driver cache, or the host cache? yes, I think taking snapshot don't flush datas. rbd devs said me that we need to freeze the filesystem with qemu-ga if we

Re: [pve-devel] create snapshots

2012-08-31 Thread Dietmar Maurer
if we don't save vm_state, we need to freeze the filesystem. (qemu-ga call xfs_freeze, which works for ext3, ext4, GFS2, XFS, and BTRFS) Currently, I'm using directsync or writethrough cache mode to be sure that the datas are always on disk. (I have a fast nvram journal in my nexenta

Re: [pve-devel] ahci: properly reset PxCMD on HBA reset - patch

2012-08-31 Thread Dietmar Maurer
Yes. But seems they are still working on that patch. -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre DERUMIER Sent: Freitag, 31. August 2012 14:05 To: pve-devel@pve.proxmox.com Subject: [pve-devel] ahci:

Re: [pve-devel] private virtual bridge

2012-09-02 Thread Dietmar Maurer
You can create as many bridges as you want (not sure why the call that private?) -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Stefan Priebe Sent: Samstag, 01. September 2012 21:00 To: pve-devel@pve.proxmox.com

Re: [pve-devel] [PATCH 3/3] livemigrate : activate xbzrle cache

2012-09-04 Thread Dietmar Maurer
after which time have you seen the endless running xbzrle? Did the remaining bytes size raise again in your case? yes. I guess we should test again when final 1.2 is out. ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] [PATCH] add qemu guest agent client

2012-09-04 Thread Dietmar Maurer
Documentation states that we can do all those thing via qmp in future! So why do we want to add all that code now? -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre Derumier Sent: Dienstag, 04. September 2012

Re: [pve-devel] [PATCH] add qemu guest agent client

2012-09-04 Thread Dietmar Maurer
chardev: backend qga_proxy not found so it's not implemented yet. (I don't know if it's on the qemu roamap, I didn't find any info) So Do you need to implemented the current procotol ? (We need guest agent to freeze filesystem for snapshots I just want to avoid unnecessary work. So please

Re: [pve-devel] create snapshots

2012-09-04 Thread Dietmar Maurer
what is the statefile argument ? I thinked that the state was saved inside the (first) disk. (so vmstate is snapshoted and we can rollback it easily) Or it's for doing some tests ? The idea is to save the state into an external file. That will work for all storage types. I don't know if the

[pve-devel] New iSCSI Multipath howto

2012-09-05 Thread Dietmar Maurer
Hi all, we just updated the wiki page: http://pve.proxmox.com/wiki/ISCSI_Multipath Please review. Comments are appreciated ;-) - Dietmar ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PATCH] add qemu guest agent client

2012-09-05 Thread Dietmar Maurer
Subject: [pve-devel] [PATCH] add qemu guest agent client This implement qemu guest agent client. I Have take the qmp client code and try to adapt it. Isn't there a way to reuse the qmp code? What exactly is the difference between QMP and QGA protocol? Can't we use a common base class?

Re: [pve-devel] New iSCSI Multipath howto

2012-09-05 Thread Dietmar Maurer
path_checker : Possible values include readsector0 and tur, emc_clariion, hp_sw, and directio. I'm using directio with nexenta. default is 'directio', so that is OK. Also, for blacklist, maybe can we add an exemple to blacklist local sda,sdb,... (If we want multipath for our san, but

Re: [pve-devel] create snapshots

2012-09-05 Thread Dietmar Maurer
I just upload a fix - please can you test again? https://git.proxmox.com/?p=pve-qemu-kvm.git;a=commitdiff;h=e3a988892faa026b88b8ceb61897b7a0753c76a9 -Original Message- From: Alexandre DERUMIER [mailto:aderum...@odiso.com] Sent: Mittwoch, 05. September 2012 09:17 To: Dietmar Maurer

Re: [pve-devel] [PATCH 2/7] add qm snapshot command

2012-09-06 Thread Dietmar Maurer
+ #need to implement config change with snapshots info + PVE::QemuServer::update_config_nolock($vmid, $conf, 1); I can work on that if you want? ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] qemu 1.2 : lsi + scsi-block don't boot

2012-09-07 Thread Dietmar Maurer
But maybe can we use scsi-hd default for lsi when we need to boot? (virtio-scsi + scsi-block no problem) We can always use that - no problem for me. ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] FVD

2012-09-07 Thread Dietmar Maurer
It was never in qemu upstream. qemu dev prefer to not maintain to much disk format . (FVD seem to be great on the paper) I just wrote a mail to the developer - maybe I will get a reply. They prefer to take idea from FVD and push them in the (futur) qcow2 v3 Yes, but qcow3 is also

Re: [pve-devel] FVD

2012-09-07 Thread Dietmar Maurer
Many thanks for the link. No idea where it went, but the qemu-devel mailing list has the patches for FVD: http://lists.gnu.org/archive/html/qemu-devel/2011-01/msg01948.html ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] FVD

2012-09-08 Thread Dietmar Maurer
It could be fine to bench FVD vs current qcow2 (with metadata preallocation, and the new lazycount option) Yes, I will try to get the latest FVD code from the developer (already got an answer from him). Will keep you informed. ___ pve-devel

Re: [pve-devel] API to get mac and vm id

2012-09-12 Thread Dietmar Maurer
I would like to create a small tool which will act as a dhcp addon to proxmox and simulate a dhcp server with perl. How should that work exactly? For this I need a way to get all vm ids and their nic's through the proxmox api. Is there a good starting example how to use the API for this?

Re: [pve-devel] API to get mac and vm id

2012-09-12 Thread Dietmar Maurer
-Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre DERUMIER Sent: Donnerstag, 13. September 2012 05:29 To: Stefan Priebe - Profihost AG Cc: pve-devel@pve.proxmox.com Subject: Re: [pve-devel] API to get mac

Re: [pve-devel] API to get mac and vm id

2012-09-12 Thread Dietmar Maurer
I would like to create a small tool which will act as a dhcp addon to proxmox and simulate a dhcp server with perl. Great ! I would like to have something like this too :) OK, can someone explain how to 'simulate a dhcp server with perl'? (Why not use 'dnsmasq' - or a real dhcp server)?

Re: [pve-devel] API to get mac and vm id

2012-09-13 Thread Dietmar Maurer
I'm reading libvirt code, they seem to use dnsmasq on each bridge Ins't that dangerous, because most times there is already an (external) DHCP server listening? ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] Snapshot Tests

2012-09-13 Thread Dietmar Maurer
i started to test the snapshot functionality using nexenta storage. But when i want to take a snapshot i get: TASK ERROR: can't snapshot volume 'Cloud1Stor1:vm-103-disk-1' Im currently try to get it work with 'qcow2' files. Tests with other storage types will follow next week.

Re: [pve-devel] qcow2 bench and speed optimisations

2012-09-13 Thread Dietmar Maurer
4)qemu-img create -f qcow2 -o size=100,preallocation=metadata,compat=1.1,lazy_refcounts=on file.img So I think it should be great to add an option for preallocatting metadatas, the boost is really huge. Above command only pre-allocates metadata, so creation is still fast? If so, we

Re: [pve-devel] Snapshot Tests

2012-09-13 Thread Dietmar Maurer
Dietmar, can we add some property to a storage plugin and check it ? (I really don't like theses hardcoded values) I just added that to prevent over-motivated hackers to run untested code ;-) The questions is how to implement that cleanly? sub storage_has_feature(feature) { ... } If you add

Re: [pve-devel] Snapshot Tests

2012-09-14 Thread Dietmar Maurer
Features when can add : snapshot,savevmstate No, ' savevmstate' is not needed (as already explained, we need to fix our qemu patch instead) ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] qcow2 bench and speed optimisations

2012-09-14 Thread Dietmar Maurer
so raw and qcow2 with metadatas alloc are sparse files, but you need a filesystem which support it. Do you know a filesystem which does not support sparse files? ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] [PATCH] added sorter to bridge selection

2012-09-14 Thread Dietmar Maurer
applied, thanks! -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of tgrodzinski Sent: Freitag, 14. September 2012 13:47 To: pve-devel@pve.proxmox.com Subject: [pve-devel] [PATCH] added sorter to bridge selection

Re: [pve-devel] about nexentaplugin eval

2012-09-17 Thread Dietmar Maurer
eval are here, if something goes wrong at last step, nexenta_add_lun_mapping_entry, and ifyou retry to create disk again, if no eval, it will hang at step1. How can I reproduce that bug? For me it work without eval. The old code was simply buggy (you can create volume multiple time

Re: [pve-devel] about nexentaplugin eval

2012-09-17 Thread Dietmar Maurer
The old code was simply buggy (you can create volume multiple time without error for example). you can't create a volume twice (nexenta side). Last step don't have eval, so It's die if you try to create a disk twice. But we also want to get an error message if we do that - old code does not

[pve-devel] nexent questions

2012-09-17 Thread Dietmar Maurer
We currently pass the following url to kvm: iscsi:$portal/$target/lun but the nexenta server does not always expose iscsi on '$portal'. Instead, there are target groups. Any idea how to fix that? Also, libiscsi does not support multipath, so maybe we are better off not using it? Can we use

Re: [pve-devel] nexent questions

2012-09-17 Thread Dietmar Maurer
We currently pass the following url to kvm: iscsi:$portal/$target/lun but the nexenta server does not always expose iscsi on ‘$portal’. Instead, there are target groups. Any idea how to fix that? maybe can we replace the portal option by management ip. and add target ip option ? no, we

Re: [pve-devel] Snapshot Tests

2012-09-18 Thread Dietmar Maurer
OK first problem is the QemuServer.pm and the hardcoded cases where snapshot is possible: die can't snapshot volume '$volid'\n if !(($scfg-{path} $volname =~ m/\.qcow2$/) || ($scfg-{type} eq 'rbd') ||

Re: [pve-devel] nexent questions

2012-09-18 Thread Dietmar Maurer
Can we use standard open-iscsi + nexenta api? Indeed, with libiscsi, you need to do multipath inside the guest. how should that work? I have attached an my old (crappy) nexenta open-iscsi plugin ;) Thanks. Another problem - we always use: target_group = All I guess that will not fit

Re: [pve-devel] Snapshot Tests

2012-09-18 Thread Dietmar Maurer
I'll try to do complete tests with all storage this week. (I'm a bit busy) Is it already possible to use the gui ? yes. ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Snapshot Tests

2012-09-18 Thread Dietmar Maurer
near the VM Backup panel -Original Message- From: Alexandre DERUMIER [mailto:aderum...@odiso.com] Sent: Dienstag, 18. September 2012 13:21 To: Dietmar Maurer Cc: pve-devel@pve.proxmox.com; Stefan Priebe Subject: Re: [pve-devel] Snapshot Tests where is the snapshot button/panel

Re: [pve-devel] corosync/pmxcfs crash (same node that previous problem)

2012-09-18 Thread Dietmar Maurer
c0f42e c0f430 c0f432 Sep 17 10:19:56 corosync [TOTEM ] Retransmit List: c0f42c c0f42d c0f42e c0f430 c0f432 Sep 17 10:19:57 corosync [TOTEM ] Retransmit List: c0f42c c0f42d c0f42e c0f430 c0f432 Sep 17 10:19:57 corosync [TOTEM ] FAILED TO RECEIVE Oh yes - this is always the same problem. But

[pve-devel] nexenta rollback problems.

2012-09-18 Thread Dietmar Maurer
Nexenta rollback seems to only work with delete_lu/create_lu. But this changes the GUID of the volume. Any idea how to avoid that? ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] nexenta rollback problems.

2012-09-18 Thread Dietmar Maurer
I guess that cause serious trouble when used with open-iscsis/dm-multipath. You already observer problems with that? I have already look at it but I never find a way to reassign the previous guid... - Mail original - De: Dietmar Maurer diet...@proxmox.com À: Alexandre DERUMIER

Re: [pve-devel] nexenta rollback problems.

2012-09-18 Thread Dietmar Maurer
I guess that cause serious trouble when used with open-iscsis/dm-multipath. You already observer problems with that? s/observer/observed/ ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] nexenta rollback problems.

2012-09-18 Thread Dietmar Maurer
Yes, indeed. that's really a pain. (BTW, I don't use multipath anymore, only lacp, take too much cpu with a lot of luns). I really tried hard, but my final conclusion is that open-iscsi/dm-multipath does not work well with dynamically allocated luns (like we do in the nexenta plugin). I know,

Re: [pve-devel] nexenta rollback problems.

2012-09-18 Thread Dietmar Maurer
So I will give up now, and we simply use libiscsi for the nexenta plugin for now. Btw, do you have any idea how we can access snapshot data for backup purposes? ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] [PATCH] first bridge device is selected by default

2012-09-18 Thread Dietmar Maurer
Subject: Re: [pve-devel] [PATCH] first bridge device is selected by default Am 14.09.2012 13:58, schrieb tgrodzinski: This patch selects the first bridge device by default. Any thoughts / comments / suggestions ? If you write a patch, It is always a good idea to explain why the patch is

Re: [pve-devel] rollback timeout

2012-09-18 Thread Dietmar Maurer
I'm just testing rollback of a qcow2 # time qm rollback 115 test2 command '/usr/bin/qemu-img snapshot -a test2 /var/lib/vz/images/115/vm- 115-disk-1.qcow2' failed: got timeout Well, you sent the patch which implements that. I have no idea why you set such short timeout on all snapshot

Re: [pve-devel] nexenta rollback problems.

2012-09-19 Thread Dietmar Maurer
Subject: Re: nexenta rollback problems. I have already look at it but I never find a way to reassign the previous guid... Seems import_lu does the trick. Do you think that will work? sub nexenta_import_lu { my ($scfg, $zvol) = @_; nexenta_request($scfg, 'import_lu', 'scsidisk',

Re: [pve-devel] nexenta rollback problems.

2012-09-19 Thread Dietmar Maurer
I'll tell the question to nexenta support. I will commit the import_lu() code for testing now. ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] rollback timeout

2012-09-19 Thread Dietmar Maurer
# time qm rollback 115 test2 command '/usr/bin/qemu-img snapshot -a test2 /var/lib/vz/images/115/vm- 115-disk-1.qcow2' failed: got timeout I set the timeout to 30 seconds now - this should be long enough. ___ pve-devel mailing list

Re: [pve-devel] snapshot gui, wrong Now location after rollback

2012-09-19 Thread Dietmar Maurer
btw, there is a bug in the '-loadstate' code - trying to fix that now. -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre DERUMIER Sent: Mittwoch, 19. September 2012 11:45 To: pve-devel@pve.proxmox.com

Re: [pve-devel] snapshot gui, wrong Now location after rollback

2012-09-19 Thread Dietmar Maurer
sorry - fixed -Original Message- From: Alexandre DERUMIER [mailto:aderum...@odiso.com] Sent: Mittwoch, 19. September 2012 12:23 To: Dietmar Maurer Cc: pve-devel@pve.proxmox.com Subject: Re: [pve-devel] snapshot gui, wrong Now location after rollback # make perl -I. ./qm

Re: [pve-devel] some qcow2 and rbd snapshots tests

2012-09-19 Thread Dietmar Maurer
rbd : snapshot works fine, with vmstate too. but rollback with vmstate , give timeout after rollback when vm is starting. Maybe loadvm doesn't work, I don't know if the vmstate is correctly save or not. vmstate can be large, so I guess I need to give longer timeout for load. Also save can

Re: [pve-devel] nexenta rollback problems.

2012-09-19 Thread Dietmar Maurer
Your case # 00014435: commstar : keep the disk guid after unmap/remap a lun (sa-api) has been updated as follows: Alexandre, There is no way to do this via SA-API . The only way to do it is via bash you as mentioned . Our current code use delete_lu/import_lu, and that does not change

Re: [pve-devel] [PATCH] qcow2 : preallocate metadatas at file creation

2012-09-19 Thread Dietmar Maurer
applied, thanks -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre Derumier Sent: Mittwoch, 19. September 2012 12:20 To: pve-devel@pve.proxmox.com Subject: [pve-devel] [PATCH] qcow2 : preallocate metadatas at

Re: [pve-devel] nexenta rollback problems.

2012-09-19 Thread Dietmar Maurer
oh, do you tried import_lu ? is it working ? I works for me, so I already committed that code. ___ pve-devel mailing list pve-devel@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PATCH] don't set timeout when loading vmstate

2012-09-20 Thread Dietmar Maurer
that is already in git -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre Derumier Sent: Donnerstag, 20. September 2012 09:06 To: pve-devel@pve.proxmox.com Subject: [pve-devel] [PATCH] don't set timeout when

Re: [pve-devel] spice news

2012-09-22 Thread Dietmar Maurer
and spice html5 seem to have a big update too http://cgit.freedesktop.org/spice/spice-html5/ I'll try to do some tests next week Thanks for the update. ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] Fwd: [sheepdog] [ANNOUNCE] Sheepdog 0.5.0 released

2012-09-23 Thread Dietmar Maurer
Subject: [pve-devel] Fwd: [sheepdog] [ANNOUNCE] Sheepdog 0.5.0 released I think that you will like - manual object recovery ;) No, I do not like the way they implemented that. IMHO It's just quick hack. But better than before ;-) ___ pve-devel

Re: [pve-devel] snapshot improvements

2012-09-24 Thread Dietmar Maurer
How is it possible to save the vmstate if the vm is not paused ? (with a lot a memory write access by example ?) I currently allocate am image equal to the size of the VM memory. Then I simply do an incremental state save (like a vm migration), and keep the VM running until (saved_bytes +

Re: [pve-devel] snapshot improvements

2012-09-24 Thread Dietmar Maurer
do you think image = size of memory if enough ? (if we have some bytes more with incremental ?) PVE/QemuServer.pm: my $driver_state_size = 32; # assume 32MB is enough to safe all driver state; my $size = $conf-{memory} + $driver_state_size; Also, any problem if some datas in

Re: [pve-devel] snapshot improvements

2012-09-25 Thread Dietmar Maurer
do you think image = size of memory if enough ? (if we have some bytes more with incremental ?) PVE/QemuServer.pm: my $driver_state_size = 32; # assume 32MB is enough to safe all driver state; my $size = $conf-{memory} + $driver_state_size; Unfortunately, the qemu code seem to

Re: [pve-devel] [PATCH] bump to 0.5

2012-09-25 Thread Dietmar Maurer
applied, thanks -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre Derumier Sent: Montag, 24. September 2012 13:34 To: pve-devel@pve.proxmox.com Subject: [pve-devel] [PATCH] bump to 0.5 Signed-off-by:

Re: [pve-devel] snapshot improvements

2012-09-26 Thread Dietmar Maurer
I also tried todo backups with snapshots but this doesn't seem to work: INFO: starting new backup job: vzdump 100 --remove 0 --mode snapshot -- compress lzo --storage backuplocal --node serv121 INFO: Starting Backup of VM 100 (qemu) INFO: status = running ERROR: Backup of VM 100 failed -

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
I also don't think that all storage can export datas with snapshots inside. rbd can export an image from a snapshot, nexenta and sheepdog too. How can you access snapshot data with nexenta? ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
The idea is that we do not backup any snapshot data. The vzdump would only include the data of from the running instance. I guess that is OK? But isn't the correct way to make a snapshot and then compress the snapshot? How can vzdump verify that the data integrity is fine? Sorry, I do not

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
I think I have already respond to that ;) 2 ways : - clone the snapshot and export image through iscsi and backup it - use zfs send through ssh (zfs send image1@snap1 /imagefile) Both ways are clumsy. Also, nexenta snapshot support is quite unusable, because you can't delete

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
The idea is that we do not backup any snapshot data. The vzdump would only include the data of from the running instance. I guess that is OK? But isn't the correct way to make a snapshot and then compress the snapshot? How can vzdump verify that the data integrity is fine? I guess I found

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
ah OK but wouldn't it be nice to be able to backup live snapshots? compress and store them on a seperate NFS server? Also Proxmox only allows to schedule beckups not to schedule snapshots ;-) Sure, that would be nice. I already have very detailed plans how to do that (but no budget).

Re: [pve-devel] [PATCH] pve-bridge : use mtu of the bridge for tap interface

2012-09-27 Thread Dietmar Maurer
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm index 157b4a9..aaf2e4a 100644 --- a/PVE/QemuServer.pm +++ b/PVE/QemuServer.pm @@ -3718,6 +3718,7 @@ my $snapshot_prepare = sub { if !(($scfg-{path} $volname =~ m/\.qcow2$/) || ($scfg-{type} eq

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
and export image through iscsi and backup it - use zfs send through ssh (zfs send image1@snap1 /imagefile) - Mail original - De: Dietmar Maurer diet...@proxmox.com À: Alexandre DERUMIER aderum...@odiso.com Cc: pve-devel@pve.proxmox.com, Stefan Priebe s.pri...@profihost.ag

Re: [pve-devel] [PATCH] pve-bridge : use mtu of the bridge for tap interface

2012-09-27 Thread Dietmar Maurer
original - De: Dietmar Maurer diet...@proxmox.com À: Alexandre Derumier aderum...@odiso.com, pve- de...@pve.proxmox.com Envoyé: Jeudi 27 Septembre 2012 10:37:17 Objet: RE: [pve-devel] [PATCH] pve-bridge : use mtu of the bridge for tap interface diff --git a/PVE/QemuServer.pm b/PVE

Re: [pve-devel] snapshot improvements

2012-09-27 Thread Dietmar Maurer
Subject: Re: [pve-devel] snapshot improvements do you mean : backup the main image then backup each snasphot increment ? Yes, something like that. zfs send can do incremental backup with zfs send -I http://docs.oracle.com/cd/E19082-01/817-2271/gfwqb/index.html Ah, good to know.

Re: [pve-devel] snapshot improvements

2012-09-28 Thread Dietmar Maurer
what do you mean by snapshot used by another snapshot? ex: image-snap1-snap2-you are here you can delete snap1 without any problem, and without need to merge. Really? - last time I tried I got 'snapshot in use' (or something like that). ___

Re: [pve-devel] pve-qemu-kvm : bump librbd to 0.52

2012-09-28 Thread Dietmar Maurer
I will not add development packages (only stable releases). -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre Derumier Sent: Freitag, 28. September 2012 12:11 To: pve-devel@pve.proxmox.com Subject:

Re: [pve-devel] snapshot improvements

2012-09-28 Thread Dietmar Maurer
maybe do you have cloned it ? no - I need to test that again. - Mail original - De: Dietmar Maurer diet...@proxmox.com À: Alexandre DERUMIER aderum...@odiso.com Cc: pve-devel@pve.proxmox.com, Stefan Priebe s.pri...@profihost.ag Envoyé: Vendredi 28 Septembre 2012 11:32:01

Re: [pve-devel] pve-qemu-kvm : bump librbd to 0.52

2012-09-28 Thread Dietmar Maurer
Subject: Re: [pve-devel] pve-qemu-kvm : bump librbd to 0.52 ok, no problem. next stable should be 0.54. Yes. I will upload that as soon as they release 0.54 ;-) ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] [PATCH] pve-bridge : use mtu of the bridge for tap interface

2012-09-28 Thread Dietmar Maurer
applied, thanks. -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel- boun...@pve.proxmox.com] On Behalf Of Alexandre Derumier Sent: Freitag, 28. September 2012 10:58 To: pve-devel@pve.proxmox.com Subject: [pve-devel] [PATCH] pve-bridge : use mtu of the

Re: [pve-devel] snapshot improvements

2012-09-29 Thread Dietmar Maurer
Are your aware of the fact that ssh needs to encrypt/decrypt all data. This needs much CPU power and is slow (still no AES support in libs). Oh we could also use netcat instead. How exactly? ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] [PATCH] volume_snapshot : add testfeature option param

2012-09-30 Thread Dietmar Maurer
This hide knowledge inside a function. Instead. I would prefer to have a declarative way specify such features. What exactly is the advantage of this patch compared with current implementation? -Original Message- From: pve-devel-boun...@pve.proxmox.com [mailto:pve-devel-

Re: [pve-devel] pve-spice 0.12 package + report

2012-10-01 Thread Dietmar Maurer
Maybe can we begin to build qemu-kvm with spice, now that 0.12 stable is out ? It'll be more simple to test/debug it. Roadmap is still the same: 1.) implement pve auth with spice 2.) implement proxy code for spice Then we can start to build qemu-kvm with spice support. IMHO it does not make

Re: [pve-devel] pve-spice 0.12 package + report

2012-10-01 Thread Dietmar Maurer
for pve-auth ? spicec client only send the password without login, I don't see how we can do this without hacking the client... So how is that expected to work? Authentication needs a user name, else it does not make much sense? ___ pve-devel

Re: [pve-devel] RFC: vm cloning implementation proposal

2012-10-02 Thread Dietmar Maurer
2 types Storage which support cloning: 1)-sheepdog,rbd (=0.52),nexenta : - the clone is made from a snapshot - can be done when sourcevm is online - we can use source vm after cloning - we can't delete a snapshot if a clone is made from the snapshot - we can't delete the

Re: [pve-devel] RFC: vm cloning implementation proposal

2012-10-03 Thread Dietmar Maurer
But first, why do I want to clone a VM? Please can you elaborate on the usage scenario first? Cloning vm is a current scenario in enterprise, generally, you have a basevm you want to duplicate in 1 second. Other scenario can be vdi desktop, you can clone in live a vm if a user need a

Re: [pve-devel] RFC: vm cloning implementation proposal

2012-10-03 Thread Dietmar Maurer
like a linux image or windows image, you want to do security updates or install new packages on the image, then make a new snapshot, and clone from this new snapshot. You want to clone a specific snapshot? Live cloning can be also good to duplicate a vm to do tests in parallel. So, I

Re: [pve-devel] RFC: vm cloning implementation proposal

2012-10-03 Thread Dietmar Maurer
You want to clone a specific snapshot? Yes. we need to clone from a snapshot for sheepdog,rbd,nexenta. So choosing the snapshot we want is good. I do not understand why we need that? Please can you explain? Live cloning include VM state? I don't think we need to clone the vmstate, But we

Re: [pve-devel] new ceph 0.48.2 stable packages

2012-10-03 Thread Dietmar Maurer
Subject: [pve-devel] new ceph 0.48.2 stable packages can you update the proxmox repo ? I already uploaded that to pvetest. Plan is to release with 2.2 ___ pve-devel mailing list pve-devel@pve.proxmox.com

Re: [pve-devel] RFC: vm cloning implementation proposal

2012-10-03 Thread Dietmar Maurer
I do not understand why we need that? Please can you explain? with rbd,nexenta,sheepdog, it's mandatory to clone from a snapshot. (that's why the sourcevm don't need to be readonly after the clone, like qcow2) Ah, I see (But it is unrelated to our snapshot implementation inside

Re: [pve-devel] RFC: vm cloning implementation proposal

2012-10-03 Thread Dietmar Maurer
Also, that 2-step approach makes implementation straight forward: 1.) create template: create a read-only snapshot of all used VM disks nexenta, rdb and sheepdog: use storage snapshot feature nfs, directory: we simply make a copy of the image 2.) clone template: make a writable copy of

  1   2   3   4   5   6   7   8   9   10   >