Re: [sheepdog] [PATCH v4] sheepdog: selectable object size support

2015-02-10 Thread Liu Yuan
On Tue, Feb 10, 2015 at 05:22:02PM +0900, Teruaki Ishizaki wrote: (2015/02/10 12:10), Liu Yuan wrote: On Tue, Jan 27, 2015 at 05:35:27PM +0900, Teruaki Ishizaki wrote: Previously, qemu block driver of sheepdog used hard-coded VDI object size. This patch enables users to handle block_size_shift

Re: [sheepdog] [PATCH] sheep: increase buffer while in cinfo collection

2015-02-10 Thread Liu Yuan
On Mon, Feb 09, 2015 at 04:00:32PM +0800, Liu Yuan wrote: Even with a 3 node cluster and only 1 vdi, killing one node and joining it back the error message that buffer too small will pop out twice, this means 512 is too small even for a single VDI cluster. Making it as 4K will mitigat

Re: [sheepdog] [PATCH] dog: type cast miss at vdi_show_progress

2015-02-10 Thread Saeki Masaki
On 2015/02/10 16:33, Liu Yuan wrote: On Tue, Feb 10, 2015 at 03:03:56PM +0900, Saeki Masaki wrote: The result of the multiplication of uint32 and uint64 are rounded to uint32 So, Progress of the command is displayed back to 0%, when it exceeds the boundaries of 32bit. This is only a display

[sheepdog] [PATCH] dog: fix to calculate a resizable max VDI size appropriately

2015-02-10 Thread Teruaki Ishizaki
A resizable max VDI size was fixed value, 4TB. So, when block_size_shift was specified more than 22, resizing VDI size over 4TB caused error. This patch enables to calculate a resizable max VDI properly. Signed-off-by: Teruaki Ishizaki ishizaki.teru...@lab.ntt.co.jp --- dog/vdi.c | 24

Re: [sheepdog] [PATCH] dog: fix to calculate a resizable max VDI size appropriately

2015-02-10 Thread Liu Yuan
On Tue, Feb 10, 2015 at 05:53:44PM +0900, Teruaki Ishizaki wrote: A resizable max VDI size was fixed value, 4TB. Why we max VDI size is 4TB if we have object size more than 4MB? I remember in the past max size is 4TB because object is 4MB. Thanks Yuan -- sheepdog mailing list

Re: [sheepdog] [PATCH v4] sheepdog: selectable object size support

2015-02-10 Thread Teruaki Ishizaki
(2015/02/10 17:58), Liu Yuan wrote: On Tue, Feb 10, 2015 at 05:22:02PM +0900, Teruaki Ishizaki wrote: (2015/02/10 12:10), Liu Yuan wrote: On Tue, Jan 27, 2015 at 05:35:27PM +0900, Teruaki Ishizaki wrote: Previously, qemu block driver of sheepdog used hard-coded VDI object size. This patch

Re: [sheepdog] [PATCH] dog: fix to calculate a resizable max VDI size appropriately

2015-02-10 Thread Teruaki Ishizaki
(2015/02/10 18:01), Liu Yuan wrote: On Tue, Feb 10, 2015 at 05:53:44PM +0900, Teruaki Ishizaki wrote: A resizable max VDI size was fixed value, 4TB. Why we max VDI size is 4TB if we have object size more than 4MB? I remember in the past max size is 4TB because object is 4MB. Sorry for

Re: [sheepdog] [PATCH] dog: type cast miss at vdi_show_progress

2015-02-10 Thread Liu Yuan
On Tue, Feb 10, 2015 at 06:22:33PM +0900, Saeki Masaki wrote: On 2015/02/10 16:33, Liu Yuan wrote: On Tue, Feb 10, 2015 at 03:03:56PM +0900, Saeki Masaki wrote: The result of the multiplication of uint32 and uint64 are rounded to uint32 So, Progress of the command is displayed back to 0%,

Re: [sheepdog] [PATCH v4] sheepdog: selectable object size support

2015-02-10 Thread Liu Yuan
On Tue, Feb 10, 2015 at 06:56:33PM +0900, Teruaki Ishizaki wrote: (2015/02/10 17:58), Liu Yuan wrote: On Tue, Feb 10, 2015 at 05:22:02PM +0900, Teruaki Ishizaki wrote: (2015/02/10 12:10), Liu Yuan wrote: On Tue, Jan 27, 2015 at 05:35:27PM +0900, Teruaki Ishizaki wrote: Previously, qemu block

Re: [sheepdog] [PATCH v4] sheepdog: selectable object size support

2015-02-10 Thread Liu Yuan
On Tue, Jan 27, 2015 at 05:35:27PM +0900, Teruaki Ishizaki wrote: Previously, qemu block driver of sheepdog used hard-coded VDI object size. This patch enables users to handle block_size_shift value for calculating VDI object size. When you start qemu, you don't need to specify additional

Re: [sheepdog] [PATCH] dog: type cast miss at vdi_show_progress

2015-02-10 Thread Saeki Masaki
On 2015/02/10 18:30, Liu Yuan wrote: On Tue, Feb 10, 2015 at 06:22:33PM +0900, Saeki Masaki wrote: On 2015/02/10 16:33, Liu Yuan wrote: On Tue, Feb 10, 2015 at 03:03:56PM +0900, Saeki Masaki wrote: The result of the multiplication of uint32 and uint64 are rounded to uint32 So, Progress of the

[sheepdog] [PATCH v2] dog: type cast miss at vdi_show_progress

2015-02-10 Thread Saeki Masaki
The result of the multiplication of uint32 and uint64 are rounded to uint32 So, Progress of the command is displayed sometimes back to 0% This is only a display issue, command processing can be completed. necessary in show_progress of calculation, change type of idx to uint64_t. However, idx in

[sheepdog] [PATCH] zookeeper: show real session timeout in zk_init()

2015-02-10 Thread Liu Yuan
From: Liu Yuan liuy...@cmss.chinamobile.com ZK server has its own session timeout range, so client's timeout value passed to server might be ignored and return the value it thinks proper. From the documentation, the server range is [tickTime * 2, tickTime * 20] e.g, if tickTime is 2s, then the

Re: [sheepdog] [PATCH v2] dog: type cast miss at vdi_show_progress

2015-02-10 Thread Liu Yuan
On Tue, Feb 10, 2015 at 08:34:44PM +0900, Saeki Masaki wrote: The result of the multiplication of uint32 and uint64 are rounded to uint32 So, Progress of the command is displayed sometimes back to 0% This is only a display issue, command processing can be completed. necessary in

[sheepdog] Build failed in Jenkins: sheepdog-build #630

2015-02-10 Thread sheepdog-jenkins
See http://jenkins.sheepdog-project.org:8080/job/sheepdog-build/630/changes Changes: [liuyuan] deb: enable dh_systemd and provide a Debian-specific service file. [mitake.hitoshi] dog: fix vdi_alter_copy for using a right block_size_shift -- [...truncated