)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"b520ea73cecaa5690c0430c50d5ee5373e7fa7b6","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_389a7262","line":11,"range":{"start_line":11,"start_character":0,"end_line":11,"end_character":36},"updated":"2020-04-17 12:46:54.000000000","message":"If there is no cyborg service deployed then there is no way to create a server with accel:device_profile in the flavor. The server create fails without even a build request is created so there will be no instance in ERROR state to be deleted.\n\nhttp://paste.openstack.org/show/792324/","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"b520ea73cecaa5690c0430c50d5ee5373e7fa7b6","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_98c10632","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"updated":"2020-04-17 12:46:54.000000000","message":"So an instance is created with accelerator first then the whole cyborg service removed from the deployment. I think the admin who decides to remove the cyborg service first needs to make sure nobody is using it, hence the instance should be deleted before the cyborg service is removed.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"b520ea73cecaa5690c0430c50d5ee5373e7fa7b6","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_f88aca18","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":49},"updated":"2020-04-17 12:46:54.000000000","message":"deployed","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":26458,"name":"Brin Zhang","email":"zhangbailin@inspur.com","username":"zhangbailin"},"change_message_id":"a067fc5ad4ae22fb379f103dc4810591568d2279","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"1f493fa4_2492b3f7","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_245fc872","updated":"2020-04-21 00:29:00.000000000","message":"yes, that not same, we can ignore this fix.\n\nBut, @Dan, gibi and I talked in IRC, from this talking we can know, two things:\n\n1.Whether the resource service removed or down(can not up by anyway), we can occupy the vGPU or FPGA etc., because we cannot use it since we lost the Cyborg service in it\u0027s region.\n\n2.nova-cinder interaction, when you want to delete the instance when the cinder service removed or down, that you can delete instance successfully, and leaked its attachments in cinder side. But nova-neutron interaction, if the neutron service removed ro down, that will be set ERROR for the deleting instance.\n\nAbove all, nova-cyborg or nova-neutron interaction scenario, if the cyborg or neutron service hit *ERROR*, if you want to delete the instance, that will be get an ERROR instance, I think the things to watch out that we cannot free Nova resources, do we need to consider these occupied Nova resources?","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"caeecf309f3e63ee5826e471f3d7f94d1172eb8d","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_bd0bf0d3","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_408acebe","updated":"2020-04-20 10:56:34.000000000","message":"\u003e Agree with Brin. So, should we handle this exception same as cinder\n \u003e to delete instance anyway or same as neutron to terminate delete\n \u003e instance? Deployed before and remove later, this scenario can be\n \u003e handled by remove all cyborg resources first. But what should we do\n \u003e if the cyborg service is abnormal?\n\nI would not add code to nova to support a faulty (\"abnormal\") cyborg service. I would fix the cyborg service instead.\n\nIf you mean by \"abnormal\" that there is a temporary network outage or a service restart of cyborg then in that case I as an admin prefer that the user get an error and then at a later time the user re-tries the delete and that will succeeds. I prefer this over the case when the first delete succeeds but the system leaks physical devices in cybrog.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":26458,"name":"Brin Zhang","email":"zhangbailin@inspur.com","username":"zhangbailin"},"change_message_id":"d3036f0eb675238c626a000ae4be6d1b28e83a6d","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_6c8f0b75","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_53ac7f9f","updated":"2020-04-19 05:52:53.000000000","message":"Not exactly the same.\n\nIn Cinder, if the Cinder service is abnormal, we can delete the instance and leave the volume on the Cinder side. Of course, we can manually delete the volume through \u0027cinder reset-state --state available [--attach-status detached] \u003cvolume\u003e\u0027.\n\nBut if cyborglien reports an exception here, we will not be able to delete the instance and set the instance to ERROR. Shouldn\u0027t we delete the instance (free up resources) to ensure that users can manually clean up its accelerators?\n\nThis is the scenario of cinder in my local env by shutdown cinder service, I paste the step and logs, please see http://paste.openstack.org/show/792375/","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"caeecf309f3e63ee5826e471f3d7f94d1172eb8d","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_784fb633","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_6c8f0b75","updated":"2020-04-20 10:56:34.000000000","message":"\u003e Not exactly the same.\n \u003e \n \u003e In Cinder, if the Cinder service is abnormal, we can delete the\n \u003e instance and leave the volume on the Cinder side. Of course, we can\n \u003e manually delete the volume through \u0027cinder reset-state --state\n \u003e available [--attach-status detached] \u003cvolume\u003e\u0027.\n \u003e \n \u003e But if cyborglien reports an exception here, we will not be able to\n \u003e delete the instance and set the instance to ERROR. Shouldn\u0027t we\n \u003e delete the instance (free up resources) to ensure that users can\n \u003e manually clean up its accelerators?\n \u003e \n\nThe problem is that the end user thinks that because the delete was successful therefore nothing to do, the resource are freed. But they are not as cyborg resource is leaked. Also I don\u0027t think that the end user can user the cyborg ARQ API do the cleanup later as he don\u0027t know the specific ARQ id. \n\n \u003e This is the scenario of cinder in my local env by shutdown cinder\n \u003e service, I paste the step and logs, please see http://paste.openstack.org/show/792375/\n\nI only checked neutron in this regard and it seems based on your trial that nova behaves differently regarding cinder and neutron failure.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":31412,"name":"Wenping Song","email":"songwenping@inspur.com","username":"songwenping"},"change_message_id":"fcde6a372d15497b364e7cd9203db8fc2cb576f3","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_408acebe","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_6c8f0b75","updated":"2020-04-20 02:53:42.000000000","message":"Agree with Brin. So, should we handle this exception same as cinder to delete instance anyway or same as neutron to terminate delete instance? Deployed before and remove later, this scenario can be handled by remove all cyborg resources first. But what should we do if the cyborg service is abnormal?","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":4393,"name":"Dan Smith","email":"dms@danplanet.com","username":"danms"},"change_message_id":"d139d18d3fe7577554bccf7fb1dae324b915e7e9","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_53ac7f9f","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_780eba3b","updated":"2020-04-17 13:28:19.000000000","message":"The same thing will happen with cinder. I agree this is not a scenario we need to account for.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":26458,"name":"Brin Zhang","email":"zhangbailin@inspur.com","username":"zhangbailin"},"change_message_id":"215bdcb2acb234dcbd86f8313444855c7cc9520e","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_9800c2ad","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_784fb633","updated":"2020-04-20 11:02:57.000000000","message":"\u003e The problem is that the end user thinks that because the delete was\n \u003e successful therefore nothing to do, the resource are freed. But\n \u003e they are not as cyborg resource is leaked. Also I don\u0027t think that\n \u003e the end user can user the cyborg ARQ API do the cleanup later as he\n \u003e don\u0027t know the specific ARQ id.\n\nMake sense.\n\n \u003e I only checked neutron in this regard and it seems based on your\n \u003e trial that nova behaves differently regarding cinder and neutron\n \u003e failure.\n\nYes their have different behaves.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":4393,"name":"Dan Smith","email":"dms@danplanet.com","username":"danms"},"change_message_id":"a3ec1831d34b92ba3333ebe3aa61e5d47df9d5af","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_245fc872","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_9800c2ad","updated":"2020-04-20 14:19:16.000000000","message":"Your example of cinder\u0027s reset-state isn\u0027t at all the same as what the patch claims right? If you add a cyborg reset-state, or just have the user delete the ARQs, then it makes sense for nova to ignore the 404 when it goes to delete them itself. However, AFAICT, this patch is talking about *any* failure, and specifically \"the cyborg service is REMOVED\" which is fundamentally different from just having already deleted the ARQs.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"bc889ef95738db1e662b9581ba2b5d67e383c3a7","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_183fb6d6","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_98c10632","updated":"2020-04-17 13:05:16.000000000","message":"I tried what happens today if neutron is removed from the deployment (form the keystone endpoints) while having an instance with neutron ports. The delete also fails with HTTP 500. So I tend to feel that removing cyborg from the deployment while a cyborg user still exists is invalid.\n\nhttps://review.opendev.org/#/c/720670","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"83f974bf036fdb6b27fdc74e8011d4fc0c5f8a5b","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"When our cloud platform is not deployed or the Cyborg service is removed for"},{"line_number":10,"context_line":"some reason, we should be able to delete the instance correctly."},{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_780eba3b","line":11,"range":{"start_line":11,"start_character":41,"end_line":11,"end_character":75},"in_reply_to":"3f4c43b2_98c10632","updated":"2020-04-17 13:06:01.000000000","message":"i kind of agree with that.\n\nthey should first delete all cyborg flavors\nsecond they should work with there users to remove all cyborg instance still running either by a resize or deleting the instnace. including any shelved instances.\nthird after completeing steps 1 and 2 they can then remove cyborg form the keystone catalog and finally remove the service.\n\nthe other question that does exist is if you have messed up for some reason should we provide a way to force delete the vm.\n\nthat woudl be an api change i think but i am inclinde to say now an consider this bug invalid.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"b520ea73cecaa5690c0430c50d5ee5373e7fa7b6","unresolved":false,"context_lines":[{"line_number":11,"context_line":"Today, if there is no Cyborg service, or depolyed before and removed later,"},{"line_number":12,"context_line":"if we want to cleanup the instance with \u0027accel:device_profile\u0027 in it\u0027s flavor"},{"line_number":13,"context_line":"when the server build failed, or the user want to delete an older instance"},{"line_number":14,"context_line":"contains \u0027accel:device_profile\u0027, that will be terminated the deleted if"},{"line_number":15,"context_line":"we did not handle the exception."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"Add unit test for handling the exception."},{"line_number":18,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3f4c43b2_3810b201","line":15,"range":{"start_line":14,"start_character":32,"end_line":15,"end_character":32},"updated":"2020-04-17 12:46:54.000000000","message":"the delete of the instance will fail due to unhandled exception","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"}],"nova/compute/utils.py":[{"author":{"_account_id":26458,"name":"Brin Zhang","email":"zhangbailin@inspur.com","username":"zhangbailin"},"change_message_id":"7c68093fd88292b366908e3cc1d5351b619a17c7","unresolved":false,"context_lines":[{"line_number":1556,"context_line":"    dp_name \u003d instance.flavor.extra_specs.get(\u0027accel:device_profile\u0027)"},{"line_number":1557,"context_line":"    if dp_name is None:"},{"line_number":1558,"context_line":"        return"},{"line_number":1559,"context_line":"    LOG.debug(\u0027Calling Cyborg to delete ARQs for instance %(instance)s\u0027,"},{"line_number":1560,"context_line":"              {\u0027instance\u0027: instance.uuid})"},{"line_number":1561,"context_line":"    try:"},{"line_number":1562,"context_line":"        cyclient \u003d cyborg.get_client(context)"},{"line_number":1563,"context_line":"        cyclient.delete_arqs_for_instance(instance.uuid)"}],"source_content_type":"text/x-python","patch_set":1,"id":"3f4c43b2_7fc6a49c","line":1560,"range":{"start_line":1559,"start_character":4,"end_line":1560,"end_character":42},"updated":"2020-04-17 09:27:29.000000000","message":"You also need to move this LOG to Line1562, that should print after you get_client connection successfully.","commit_id":"820befc4a9e7b1370f66879afc27d50cb8ff0cd0"},{"author":{"_account_id":9708,"name":"Balazs Gibizer","display_name":"gibi","email":"gibizer@gmail.com","username":"gibi"},"change_message_id":"b520ea73cecaa5690c0430c50d5ee5373e7fa7b6","unresolved":false,"context_lines":[{"line_number":1566,"context_line":"    except exception.AcceleratorRequestOpFailed as e:"},{"line_number":1567,"context_line":"        LOG.exception(\u0027Failed to delete accelerator requests for \u0027"},{"line_number":1568,"context_line":"                      \u0027instance %s. Exception: %s\u0027, instance.uuid, e)"},{"line_number":1569,"context_line":"    except Exception as e:"},{"line_number":1570,"context_line":"        LOG.exception(\u0027Failed to communicate with Cyborg. Exception: %s\u0027, e)"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4c43b2_78fa3a75","line":1570,"range":{"start_line":1569,"start_character":0,"end_line":1570,"end_character":76},"updated":"2020-04-17 12:46:54.000000000","message":"By this, in case of a temporary failure to communicate with cyborg we allow deleting the instance without deleting the accelerator usage in cyborg and therefore leaking accelerator devices on the cyborg side. While if we fail the delete in this case then the user can try it later.","commit_id":"6c3c90a57dbec8520b6dc3a77dd5feb6113707d8"}]}