Created attachment 1122842 [details] engine-setup log Description of problem: Recently a patch was introduced to provide user with information regarding async tasks still running in the system that prevent user from going through the upgrade, see https://bugzilla.redhat.com/show_bug.cgi?id=1290528. In terms of usability I think that the result is not idle. e.g. I tried to upgrade my setup and got the following message: The following system tasks have been found running in the system: Task ID: 5c4cfa8f-175a-4786-9381-1aa274416acb Task Name: Unknown Task Description: Unknown Started at: 30 DC Name: golden_env_mixed The following commands have been found running in the system: The following compensations have been found running in the system: This information is really not very sufficient. Of course I asked around and then looked in setup log for the task ID, which resulted with some information (action type, which is an enum from engine java code) and a hint leading me to guess that my problem was due to difference between a VM's configured and actual timezone. If I wasn't a QE guy for Ovirt that specifically tested timezone configuration recently, no way that I would deduct that this was the problem To me it doesn't make sense that the information regarding this task exist and can practically be pin pointed to the user, but still I get 'Unknown' instead. I would suggest finding a solution to provide user with full task description, or at least some sort of hint for what should be he's next step in handling the issue. Needless to say that if I were new to the product I would be utterly clueless in front of such message. Version-Release number of selected component (if applicable): upgrade from rhevm-3.6.3-0.1.el6.noarch to rhevm-3.6.3.1-0.1.el6.noarch Steps to Reproduce: 1. Have a 3.6.3-0.1.el6.noarch env. 2. Create a vm with some rhel os and guest-tools on it. 3. If guest agent reports a different time zone than configured, keep it that way, if they are the same, change configuration of the vm in engine and restart the vm, so engine will report the difference. 4. attempt to upgrade engine to rhevm-3.6.3.1-0.1.el6.noarch Actual results: During setup the following message appear and eventually you are forced to cancel the upgrade. The following system tasks have been found running in the system: Task ID: 5c4cfa8f-175a-4786-9381-1aa274416acb Task Name: Unknown Task Description: Unknown Started at: 30 DC Name: golden_env_mixed The following commands have been found running in the system: The following compensations have been found running in the system: Expected results: A more informative message that helps user solve his blocking issue. Additional info:
We need to decide on how to tell the user\engine to resolve issues like this.
There are two issues here: 1. How to give the user more meaningful information in engine-setup. See also bug 1240940 for this. If (why?) we do not want to expose this information using the existing enum, might use other means? api/sdk? Might not be needed at all if (2.) below is enough. 2. How to let the user actually handle/monitor pending tasks. We have a Tasks tab in the web admin. Do all of them appear there? Do we have good means to show progress? Do we allow killing/removing/rolling back dead/non-critical/etc ones? For some of the above we might have/need open BZs, not sure. Yaniv - how should we continue?
(In reply to Yedidyah Bar David from comment #2) > There are two issues here: > > 1. How to give the user more meaningful information in engine-setup. > See also bug 1240940 for this. If (why?) we do not want to expose this > information using the existing enum, might use other means? api/sdk? > Might not be needed at all if (2.) below is enough. > > 2. How to let the user actually handle/monitor pending tasks. > We have a Tasks tab in the web admin. > Do all of them appear there? > Do we have good means to show progress? > Do we allow killing/removing/rolling back dead/non-critical/etc ones? > > For some of the above we might have/need open BZs, not sure. > > Yaniv - how should we continue? I am less concerned about what tasks are running and more concerned about user being stuck for a unknown amount of time on a unknown task. We should make it clear to the use on what to do to resolve issues, if we can. I would consult with infra what they expect the user to do and what he needs to do if a task is stuck.
(In reply to Yaniv Dary from comment #3) > I am less concerned about what tasks are running and more concerned about > user being stuck for a unknown amount of time on a unknown task. We should > make it clear to the use on what to do to resolve issues, if we can. > I would consult with infra what they expect the user to do and what he needs > to do if a task is stuck. I wrote my above message after briefly discussing this with Oved. He said users should use the existing Tasks tab. Sefi - Can you please reproduce and check if you see there a task?
Sorry for the late reply. I tried to re produce the issue with latest 'rhevm-3.6.4-0.1.el6.noarch'. I made sure that TZ configuration differs from actual TZ setting on a rhel 7.2 vm with ovirt-guest-agent reporting TZ. I get the exclamation mark on the vm. No task is shown in task tab. The difference is, that this time when tried to upgrade (engine-setup, although no new version is applied, let me know if that's a different flow and I'll do an "actual" upgrade), and setup did not report any running task interrupting the setup. let me know what is our next step here.
No idea. Attached log has: 2016-02-10 16:45:51 DEBUG otopi.ovirt_engine_setup.engine_common.database database.execute:171 Database: 'None', Statement: ' select async_tasks.action_type, async_tasks.task_id, async_tasks.started_at, storage_pool.name from async_tasks, storage_pool where async_tasks.storage_pool_id = storage_pool.id ', args: {} 2016-02-10 16:45:51 DEBUG otopi.ovirt_engine_setup.engine_common.database database.execute:176 Creating own connection 2016-02-10 16:45:51 DEBUG otopi.ovirt_engine_setup.engine_common.database database.execute:221 Result: [{'started_at': datetime.datetime(2016, 2, 9, 14, 18, 58, 548000, tzinfo=<psycopg2.tz.FixedOffsetTimezone object at 0x2c65e10>), 'task_id': '5c4cfa8f-175a-4786-9381-1aa274416acb', 'action_type': 44, 'name': 'golden_env_mixed'}] Oved - any reliable way to reproduce (cause a task of this type to be started, preferably in a way that will cause it to take a long time)?
You want a task that is "unknown" to the setup? You can either hack the DB, or run one that you know is unknown and takes a lot of time (storage operation I guess).
(In reply to Oved Ourfali from comment #7) > You want a task that is "unknown" to the setup? > You can either hack the DB, or run one that you know is unknown and takes a > lot of time (storage operation I guess). Since this bug was opened about type 44, I am asking about it. Of course it will be nice if we can test that all tasks that have a reasonable chance to take more than a few seconds, will appear in the Tasks tab, but let's start with current case. What's 44? How to cause starting it?
Sefi is currently handling 44 with Arik Hadas. Following a discussion with Oved: Long running tasks should appear in the Tasks tab, and engine-setup should tell user to look there for more info. I now opened bug 1320498 to require that.
Moving from 4.0 alpha to 4.0 beta since 4.0 alpha has been already released and bug is not ON_QA.
oVirt 4.0 beta has been released, moving to RC milestone.
Closing old bugs. Please reopen if still relevant. Patches are welcomed.