From: zhanghailiang <zhang.zhanghailiang@huawei.com>
To: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: lizhijian@cn.fujitsu.com, quintela@redhat.com,
yunhong.jiang@intel.com, eddie.dong@intel.com,
peter.huangpeng@huawei.com, qemu-devel@nongnu.org,
arei.gonglei@huawei.com, stefanha@redhat.com,
amit.shah@redhat.com, yanghy@cn.fujitsu.com
Subject: Re: [Qemu-devel] [PATCH COLO-Frame v9 08/32] COLO/migration: establish a new communication path from destination to source
Date: Wed, 21 Oct 2015 16:33:55 +0800 [thread overview]
Message-ID: <56274DF3.1020509@huawei.com> (raw)
In-Reply-To: <20151020193256.GD2520@work-vm>
On 2015/10/21 3:32, Dr. David Alan Gilbert wrote:
> * zhanghailiang (zhang.zhanghailiang@huawei.com) wrote:
>> On 2015/10/19 17:54, Dr. David Alan Gilbert wrote:
>>> * zhanghailiang (zhang.zhanghailiang@huawei.com) wrote:
>>>> Add a new member 'to_src_file' to MigrationIncomingState and a
>>>> new member 'from_dst_file' to MigrationState.
>>>> They will be used for returning messages from destination to source.
>>>> It will also be used by post-copy migration.
>>>>
>>>> Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
>>>> Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
>>>> Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
>>>> ---
>>>> include/migration/migration.h | 3 ++-
>>>> migration/colo.c | 43 +++++++++++++++++++++++++++++++++++++++++++
>>>> 2 files changed, 45 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/include/migration/migration.h b/include/migration/migration.h
>>>> index 6488e03..0c94103 100644
>>>> --- a/include/migration/migration.h
>>>> +++ b/include/migration/migration.h
>>>> @@ -50,7 +50,7 @@ typedef QLIST_HEAD(, LoadStateEntry) LoadStateEntry_Head;
>>>> /* State for the incoming migration */
>>>> struct MigrationIncomingState {
>>>> QEMUFile *from_src_file;
>>>> -
>>>> + QEMUFile *to_src_file;
>>>> int state;
>>>>
>>>> bool have_colo_incoming_thread;
>>>> @@ -74,6 +74,7 @@ struct MigrationState
>>>> QemuThread thread;
>>>> QEMUBH *cleanup_bh;
>>>> QEMUFile *to_dst_file;
>>>> + QEMUFile *from_dst_file;
>>>> int parameters[MIGRATION_PARAMETER_MAX];
>>>>
>>>> int state;
>>>> diff --git a/migration/colo.c b/migration/colo.c
>>>> index a341eee..5f4fb20 100644
>>>> --- a/migration/colo.c
>>>> +++ b/migration/colo.c
>>>> @@ -39,6 +39,20 @@ bool migration_incoming_in_colo_state(void)
>>>> static void *colo_thread(void *opaque)
>>>> {
>>>> MigrationState *s = opaque;
>>>> + int fd, ret = 0;
>>>> +
>>>> + /* Dup the fd of to_dst_file */
>>>> + fd = dup(qemu_get_fd(s->to_dst_file));
>>>> + if (fd == -1) {
>>>> + ret = -errno;
>>>> + goto out;
>>>> + }
>>>> + s->from_dst_file = qemu_fopen_socket(fd, "rb");
>>>
>>> In my postcopy code I add the return-path opening as a new
>>> method on QEMUFile, that way if we get a return path working
>>> on another transport (RDMA which I hope to do) then it
>>> works; have a look at 'Return path: Open a return path on QEMUFile for sockets'
>>>
>>
>> I have looked at it. That's a good solution, we use the same fd for return path, and
>> i don't have to call qemu_file_shutdown two times in failover process.
>>
>>>> + if (!s->from_dst_file) {
>>>> + ret = -EINVAL;
>>>> + error_report("Open QEMUFile failed!");
>>>
>>> In errors, try to give detail of where a problem was;
>>> e.g. 'colo_thread: Open QEMUFile from_dst failed'.
>>>
>>
>> OK. I will fix it in next version.
>>
>>>> + goto out;
>>>> + }
>>>>
>>>> qemu_mutex_lock_iothread();
>>>> vm_start();
>>>> @@ -47,9 +61,17 @@ static void *colo_thread(void *opaque)
>>>>
>>>> /*TODO: COLO checkpoint savevm loop*/
>>>>
>>>> +out:
>>>> + if (ret < 0) {
>>>> + error_report("Detect some error: %s", strerror(-ret));
>>>> + }
>>>
>>> Again, best to say where the error happened.
>>>
>>
>> Hmm, it is a little difficult to say where exactly this error happened here,
>> what i can do is to error out in the place where the error happened.
>> Here is only a summary for the error.
>
> OK, just add the function name then, e.g.
> error_report("%s: %s", __func__, strerror(-ret));
>
> it just helps when you're looking at a log file, to understand where it came
> from; especially if the person who sent you the log file didn't tell you much :-)
>
OK, that's a good idea, will fix in next version. Thanks.
>>
>>>
>>>> migrate_set_state(&s->state, MIGRATION_STATUS_COLO,
>>>> MIGRATION_STATUS_COMPLETED);
>>>>
>>>> + if (s->from_dst_file) {
>>>> + qemu_fclose(s->from_dst_file);
>>>> + }
>>>> +
>>>> qemu_mutex_lock_iothread();
>>>> qemu_bh_schedule(s->cleanup_bh);
>>>> qemu_mutex_unlock_iothread();
>>>> @@ -86,12 +108,33 @@ void colo_init_checkpointer(MigrationState *s)
>>>> void *colo_process_incoming_thread(void *opaque)
>>>> {
>>>> MigrationIncomingState *mis = opaque;
>>>> + int fd, ret = 0;
>>>>
>>>> migrate_set_state(&mis->state, MIGRATION_STATUS_ACTIVE,
>>>> MIGRATION_STATUS_COLO);
>>>>
>>>> + fd = dup(qemu_get_fd(mis->from_src_file));
>>>> + if (fd < 0) {
>>>> + ret = -errno;
>>>> + goto out;
>>>> + }
>>>> + mis->to_src_file = qemu_fopen_socket(fd, "wb");
>>>> + if (!mis->to_src_file) {
>>>> + ret = -EINVAL;
>>>> + error_report("Can't open incoming channel!");
>>>> + goto out;
>>>> + }
>>>
>>> Same as above.
>>
>> OK. Will fix it.
>>
>> Thanks,
>> zhanghailiang
>>>
>>>> /* TODO: COLO checkpoint restore loop */
>>>>
>>>> +out:
>>>> + if (ret < 0) {
>>>
>>>> + error_report("colo incoming thread will exit, detect error: %s",
>>>> + strerror(-ret));
>>>> + }
>>>> +
>>>> + if (mis->to_src_file) {
>>>> + qemu_fclose(mis->to_src_file);
>>>> + }
>>>> migration_incoming_exit_colo();
>>>>
>>>> return NULL;
>>>> --
>>>> 1.8.3.1
>>>>
>>>>
>>> --
>>> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
>>>
>>> .
>>>
>>
>>
> --
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
>
> .
>
next prev parent reply other threads:[~2015-10-21 8:36 UTC|newest]
Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-09-02 8:22 [Qemu-devel] [PATCH COLO-Frame v9 00/32] COarse-grain LOck-stepping(COLO) Virtual Machines for Non-stop Service (FT) zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 01/32] configure: Add parameter for configure to enable/disable COLO support zhanghailiang
2015-10-02 15:10 ` Dr. David Alan Gilbert
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 02/32] migration: Introduce capability 'colo' to migration zhanghailiang
2015-10-02 16:02 ` Eric Blake
2015-10-08 6:34 ` zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 03/32] COLO: migrate colo related info to slave zhanghailiang
2015-10-02 18:45 ` Dr. David Alan Gilbert
2015-10-08 6:48 ` zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 04/32] migration: Add state records for migration incoming zhanghailiang
2015-10-09 16:18 ` Dr. David Alan Gilbert
2015-10-10 7:07 ` zhanghailiang
2015-10-16 11:14 ` Dr. David Alan Gilbert
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 05/32] migration: Integrate COLO checkpoint process into migration zhanghailiang
2015-10-09 16:53 ` Dr. David Alan Gilbert
2015-10-10 6:25 ` zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 06/32] migration: Integrate COLO checkpoint process into loadvm zhanghailiang
2015-10-19 9:17 ` Dr. David Alan Gilbert
2015-10-20 8:04 ` zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 07/32] migration: Rename the'file' member of MigrationState and MigrationIncomingState zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 08/32] COLO/migration: establish a new communication path from destination to source zhanghailiang
2015-10-19 9:54 ` Dr. David Alan Gilbert
2015-10-20 8:30 ` zhanghailiang
2015-10-20 19:32 ` Dr. David Alan Gilbert
2015-10-21 8:33 ` zhanghailiang [this message]
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 09/32] COLO: Implement colo checkpoint protocol zhanghailiang
2015-10-21 12:17 ` Eric Blake
2015-10-22 7:13 ` zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 10/32] COLO: Add a new RunState RUN_STATE_COLO zhanghailiang
2015-10-21 12:18 ` Eric Blake
2015-10-22 6:58 ` zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 11/32] QEMUSizedBuffer: Introduce two help functions for qsb zhanghailiang
2015-09-02 8:22 ` [Qemu-devel] [PATCH COLO-Frame v9 12/32] COLO: Save PVM state to secondary side when do checkpoint zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 13/32] COLO: Load PVM's dirty pages into SVM's RAM cache temporarily zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 14/32] COLO: Load VMState into qsb before restore it zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 15/32] COLO: Flush PVM's cached RAM into SVM's memory zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 16/32] COLO: synchronize PVM's state to SVM periodically zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 17/32] COLO failover: Introduce a new command to trigger a failover zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 18/32] COLO failover: Introduce state to record failover process zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 19/32] COLO: Implement failover work for Primary VM zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 20/32] COLO: Implement failover work for Secondary VM zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 21/32] COLO: implement default failover treatment zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 22/32] qmp event: Add event notification for COLO error zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 23/32] COLO failover: Shutdown related socket fd when do failover zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 24/32] COLO failover: Don't do failover during loading VM's state zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 25/32] COLO: Control the checkpoint delay time by migrate-set-parameters command zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 26/32] COLO: Implement shutdown checkpoint zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 27/32] COLO: Update the global runstate after going into colo state zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 28/32] savevm: Split load vm state function qemu_loadvm_state zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 29/32] COLO: Separate the process of saving/loading ram and device state zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 30/32] COLO: Split qemu_savevm_state_begin out of checkpoint process zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 31/32] COLO: Add block replication into colo process zhanghailiang
2015-09-02 8:23 ` [Qemu-devel] [PATCH COLO-Frame v9 32/32] COLO: Add net packets treatment into COLO zhanghailiang
2015-09-02 9:03 ` [Qemu-devel] [PATCH COLO-Frame v9 00/32] COarse-grain LOck-stepping(COLO) Virtual Machines for Non-stop Service (FT) Yang Hongyang
2015-09-02 9:17 ` zhanghailiang
2015-09-09 3:36 ` zhanghailiang
2015-09-15 10:40 ` zhanghailiang
2015-10-21 14:10 ` Dr. David Alan Gilbert
2015-10-22 9:01 ` zhanghailiang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=56274DF3.1020509@huawei.com \
--to=zhang.zhanghailiang@huawei.com \
--cc=amit.shah@redhat.com \
--cc=arei.gonglei@huawei.com \
--cc=dgilbert@redhat.com \
--cc=eddie.dong@intel.com \
--cc=lizhijian@cn.fujitsu.com \
--cc=peter.huangpeng@huawei.com \
--cc=qemu-devel@nongnu.org \
--cc=quintela@redhat.com \
--cc=stefanha@redhat.com \
--cc=yanghy@cn.fujitsu.com \
--cc=yunhong.jiang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).