[opensuse-buildservice] RFC: provide reason why job may hang in state scheduled
Hi, I have implemented a feature that can show if a job hangs in state scheduled because there are no workers with the capabilities to build the job (constraints not fulfilled). We have two cases: 1. No worker at all can fulfill the constraints. Then the following detailed string is displayed: "no worker provides the capabilities to comply with the constraints (constraint missmatch hint: hardware) (constraint missmatch hint: sandbox)" 2. There is no idle worker, but a worker that is building, away or down can build this job later. At the moment the detailed string for case 2 that is displayed in osc r -v and on the mouse-over-event at build.o.o is not very significant. It just shows how many out of all workers (that are not in idle state) could build this job. See [1] # osc r -v Application:Geo sfcgal -r openSUSE_Factory_ARM openSUSE_Factory_ARM aarch64 scheduled: 159/627 building openSUSE_Factory_ARM armv7l scheduled Which means: 159 out of 627 workers in state building could build this job if they change to idle. You see. Not very self-explanatory. I would change the string in something like this: "no idle workers found for constraints: (3 in state building) (1 in state away)" But i am not sure and don't want to change the string every time we deploy. Any suggestions? Thanks and regards, Marco [1] http://paste.opensuse.org/35759685 -- Marco Strigl SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nuremberg, Germany GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg) -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org
On Donnerstag, 20. Oktober 2016, 15:23:13 CEST wrote Marco Strigl:
Hi,
I have implemented a feature that can show if a job hangs in state scheduled because there are no workers with the capabilities to build the job (constraints not fulfilled).
We have two cases:
1. No worker at all can fulfill the constraints. Then the following detailed string is displayed: "no worker provides the capabilities to comply with the constraints (constraint missmatch hint: hardware) (constraint missmatch hint: sandbox)"
It is quite long IMHO. Imagine how this looks when you do an "osc r -v" on your terminal. On the other side, it is self explaining :)
2. There is no idle worker, but a worker that is building, away or down can build this job later.
At the moment the detailed string for case 2 that is displayed in osc r -v and on the mouse-over-event at build.o.o is not very significant. It just shows how many out of all workers (that are not in idle state) could build this job.
See [1]
# osc r -v Application:Geo sfcgal -r openSUSE_Factory_ARM openSUSE_Factory_ARM aarch64 scheduled: 159/627 building openSUSE_Factory_ARM armv7l scheduled
Which means: 159 out of 627 workers in state building could build this job if they change to idle. You see. Not very self-explanatory.
I would change the string in something like this: "no idle workers found for constraints: (3 in state building) (1 in state away)"
Would be too long IMHO. But pointing to workers and constraints is important IMHO. Maybe just "159 possible workers" ?
But i am not sure and don't want to change the string every time we deploy.
Any suggestions?
Thanks and regards, Marco
-- Adrian Schroeter email: adrian@suse.de SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg) Maxfeldstraße 5 90409 Nürnberg Germany -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org
On 10/24/2016 09:30 AM, Adrian Schröter wrote:
On Donnerstag, 20. Oktober 2016, 15:23:13 CEST wrote Marco Strigl:
Hi,
I have implemented a feature that can show if a job hangs in state scheduled because there are no workers with the capabilities to build the job (constraints not fulfilled).
We have two cases:
1. No worker at all can fulfill the constraints. Then the following detailed string is displayed: "no worker provides the capabilities to comply with the constraints (constraint missmatch hint: hardware) (constraint missmatch hint: sandbox)"
It is quite long IMHO. Imagine how this looks when you do an "osc r -v" on your terminal. On the other side, it is self explaining :)
Mhm. We can shorten it a little bit: "no possible workers (constrains missmatch hints: hardware / sandbox)" This would match the suggestion below with the "xyz possible workers".
2. There is no idle worker, but a worker that is building, away or down can build this job later.
At the moment the detailed string for case 2 that is displayed in osc r -v and on the mouse-over-event at build.o.o is not very significant. It just shows how many out of all workers (that are not in idle state) could build this job.
See [1]
# osc r -v Application:Geo sfcgal -r openSUSE_Factory_ARM openSUSE_Factory_ARM aarch64 scheduled: 159/627 building openSUSE_Factory_ARM armv7l scheduled
Which means: 159 out of 627 workers in state building could build this job if they change to idle. You see. Not very self-explanatory.
I would change the string in something like this: "no idle workers found for constraints: (3 in state building) (1 in state away)"
Would be too long IMHO. But pointing to workers and constraints is important IMHO.
Maybe just "159 possible workers" ?
In the first step this information is ok. Perhaps we need some kind of detail level? Or an option to query the constraint violation details?
But i am not sure and don't want to change the string every time we deploy.
Any suggestions?
Thanks and regards, Marco
Thanks for your comments :-) -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org
On Montag, 24. Oktober 2016 09:30:07 CEST Adrian Schröter wrote:
On Donnerstag, 20. Oktober 2016, 15:23:13 CEST wrote Marco Strigl:
Hi,
I have implemented a feature that can show if a job hangs in state scheduled because there are no workers with the capabilities to build the job (constraints not fulfilled).
We have two cases:
1. No worker at all can fulfill the constraints. Then the following detailed string is displayed: "no worker provides the capabilities to comply with the constraints (constraint missmatch hint: hardware) (constraint missmatch hint: sandbox)" It is quite long IMHO. Imagine how this looks when you do an "osc r -v" on your terminal. On the other side, it is self explaining :)
"No worker compliant with constraints. Mismatches: [hardware, sandbox]"
2. There is no idle worker, but a worker that is building, away or down can build this job later.
At the moment the detailed string for case 2 that is displayed in osc r -v and on the mouse-over-event at build.o.o is not very significant. It just shows how many out of all workers (that are not in idle state) could build this job.
See [1]
# osc r -v Application:Geo sfcgal -r openSUSE_Factory_ARM openSUSE_Factory_ARM aarch64 scheduled: 159/627 building openSUSE_Factory_ARM armv7l scheduled
Which means: 159 out of 627 workers in state building could build this job if they change to idle. You see. Not very self-explanatory.
I would change the string in something like this: "no idle workers found for constraints: (3 in state building) (1 in state away)" Would be too long IMHO. But pointing to workers and constraints is important IMHO.
Maybe just "159 possible workers" ?
capable, compatible, compliant? This additional output is a very welcome addition, thanks for working on it! Kind regards, Stefan -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org
On Mon, Oct 24, 2016 at 11:35 AM, Brüns, Stefan <Stefan.Bruens@rwth-aachen.de> wrote:
1. No worker at all can fulfill the constraints. Then the following detailed string is displayed: "no worker provides the capabilities to comply with the constraints (constraint missmatch hint: hardware) (constraint missmatch hint: sandbox)" It is quite long IMHO. Imagine how this looks when you do an "osc r -v" on your terminal. On the other side, it is self explaining :)
"No worker compliant with constraints. Mismatches: [hardware, sandbox]"
Hopefully it is a rare situation. Verbosity is not a bad thing in rare error messages IMHO. But I find Stefan's verbiage both succinct and clear. Greg -- Greg Freemyer Hillary says Trump is the only Republican she can beat, why should we vote for him? Evan McMullin is surging in the west of the Rockies Deny Hillary, vote Evan McMullin for President -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org
On 10/24/2016 05:51 PM, Greg Freemyer wrote:
On Mon, Oct 24, 2016 at 11:35 AM, Brüns, Stefan <Stefan.Bruens@rwth-aachen.de> wrote:
1. No worker at all can fulfill the constraints. Then the following detailed string is displayed: "no worker provides the capabilities to comply with the constraints (constraint missmatch hint: hardware) (constraint missmatch hint: sandbox)" It is quite long IMHO. Imagine how this looks when you do an "osc r -v" on your terminal. On the other side, it is self explaining :)
"No worker compliant with constraints. Mismatches: [hardware, sandbox]"
Hopefully it is a rare situation. Verbosity is not a bad thing in rare error messages IMHO.
I think it is not very common. But if it happens the message should give the user a clue what went wrong.
But I find Stefan's verbiage both succinct and clear.
Yes. I like it too.
Greg
-- Greg Freemyer Hillary says Trump is the only Republican she can beat, why should we vote for him? Evan McMullin is surging in the west of the Rockies Deny Hillary, vote Evan McMullin for President
I will implement the following. Already checked in the version with "possible workers", but Stefan is right. Compliant sounds better. Will change it tomorrow morning. For jobs that cannot be built on any worker: "no compliant workers (constraints mismatch hint: hardware sandbox)" And for jobs that could be built later: "waiting for x compliant workers (y down)" I will go with this at first. Thanks for your thoughts, Marco -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org
participants (5)
-
Adrian Schröter -
Brüns, Stefan -
Greg Freemyer -
Marco Strigl -
Marco Strigl