New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 709557 link

Starred by 4 users

Issue metadata

Status: Fixed
Owner:
Closed: Nov 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Feature



Sign in to add a comment

Add support to "lease" a swarming task

Project Member Reported by mar...@chromium.org, Apr 7 2017

Issue description

It's a frequently requested feature but was not yet implemented because the Swarming bot doesn't know how to pass through NAT routers.

Goal:
Similar to a 'terminate' task, add a 'lease' task. This task would have a duration associated (24h default?) and could be canceled early by the user or an admin. But unlike 'terminate', the swarming bot process wouldn't terminate, it would put itself to sleep with periodic poll to determine if the lease was canceled.

Open questions:
- How a dev is supposed to access the bots. Other services open a temporary ssh server or other mechanism. We can't assume the worker is externally accessible and the Swarming server itself cannot act as a proxy.
- Security wise, the group that have access to lease bots are privileged users or create a new independent group? Likely the later.

 

Comment 1 by estaab@chromium.org, Jun 20 2017

Status: Available (was: Untriaged)

Comment 2 by mar...@chromium.org, Oct 19 2017

Labels: -Pri-3 Pri-2
A Skia specific feature request is isssue skia:7098.

There's multiple facets here, some are where Machine Provider ough to give a plain VM but not all machines are managed by MP at the moment, so either we make this a forcing function or we keep it at the Swarming level.

There are precedent, a good example is the Go CI system.

We'd probably want it at the Swarming level so that the tree would be mapped in, which in some cases, like Windows and OSX, is really important to be able to RDP in to see what happened in the UI. Having a debugger ready would be a nice to have too. I'll keep it focused on Swarming because of this. This doesn't preclude a pure MP lease Web UI.

Raising priority because this request constantly comes back.


Cc: nedngu...@google.com
Duplicate of 589982 ?

Comment 5 by mar...@chromium.org, Oct 25 2017

Cc: -nedngu...@google.com bpastene@chromium.org flowblok@chromium.org yihongg@chromium.org nednguyen@chromium.org
Summary: Add support to "lease" a swarming task (was: Add support to "lease" a swarming bot)
Issue 589982 is about grabbing a bot. I renamed this issue to be focused on "grabbing a task". I'm thinking about not implementing the "grab a bot" at all after all.

Comment 6 by mar...@chromium.org, Oct 25 2017

Cc: mar...@chromium.org
 Issue 658392  has been merged into this issue.

Comment 7 by mar...@chromium.org, Oct 25 2017

Issue 589982 has been merged into this issue.

Comment 8 by mar...@chromium.org, Oct 25 2017

Cc: kjlubick@chromium.org
Components: Infra>Platform>Swarming>WebUI
Project Member

Comment 9 by bugdroid1@chromium.org, Oct 25 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/luci/luci-py.git/+/0f778293df7bafe1f0b0b44931f65dc638113a3b

commit 0f778293df7bafe1f0b0b44931f65dc638113a3b
Author: Marc-Antoine Ruel <maruel@chromium.org>
Date: Wed Oct 25 17:54:53 2017

Add debug_task.py to trigger a debug task.

Bug:  709557 
Change-Id: Ib071f7395cec76a87cba0eded9a881d3068bba76
Reviewed-on: https://chromium-review.googlesource.com/736371
Reviewed-by: Benjamin Pastene <bpastene@chromium.org>
Commit-Queue: Marc-Antoine Ruel <maruel@chromium.org>

[add] https://crrev.com/0f778293df7bafe1f0b0b44931f65dc638113a3b/client/tools/debug_task.py

Cc: rmis...@chromium.org
Project Member

Comment 11 by bugdroid1@chromium.org, Oct 25 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/luci/luci-py.git/+/ed1fd851ab0de84771c35d7a1dd04a2d573bab1a

commit ed1fd851ab0de84771c35d7a1dd04a2d573bab1a
Author: Marc-Antoine Ruel <maruel@chromium.org>
Date: Wed Oct 25 18:21:40 2017

Fix the debug_task.py generated task url

Had forgot to confirm it was the valid format.

Bug:  709557 
Change-Id: Ida41a5fde48d63ccee8b6f00e13f7c639f9b6da6
Reviewed-on: https://chromium-review.googlesource.com/737421
Reviewed-by: Benjamin Pastene <bpastene@chromium.org>
Commit-Queue: Marc-Antoine Ruel <maruel@chromium.org>

[modify] https://crrev.com/ed1fd851ab0de84771c35d7a1dd04a2d573bab1a/client/tools/debug_task.py

Project Member

Comment 13 by bugdroid1@chromium.org, Nov 1 2017

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal/+/194aa3d0d92c58b5f5663c8e17850c0c9fd88df6

commit 194aa3d0d92c58b5f5663c8e17850c0c9fd88df6
Author: Benjamin Pastene <bpastene@chromium.org>
Date: Wed Nov 01 19:34:12 2017

\o/

Will not mark as fixed yet because of the Retry button regression.
Cc: -kjlubick@chromium.org
Owner: kjlubick@chromium.org
Status: Started (was: Available)
I can repro the retry button regression.  I'll have a fix in soon, hopefully.
Status: Fixed (was: Started)
Fix in https://chromium-review.googlesource.com/c/infra/luci/luci-py/+/758880 has been deployed to chromium-swarm(-dev)? and chrome-swarming
Project Member

Comment 17 by bugdroid1@chromium.org, Nov 9 2017

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal/+/68746c5ac2017920e99a2a77e78b984f0de53598

commit 68746c5ac2017920e99a2a77e78b984f0de53598
Author: Marc-Antoine Ruel <maruel@chromium.org>
Date: Thu Nov 09 19:38:34 2017

Sign in to add a comment