)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"333535c0f206514aa21b1602871e94a9e8a56ed4","unresolved":false,"context_lines":[{"line_number":66,"context_line":""},{"line_number":67,"context_line":"[*] https://bugzilla.redhat.com/show_bug.cgi?id\u003d1010885#c2 --"},{"line_number":68,"context_line":"    kvm_init_vcpu failed: Cannot allocate memory in NUMA"},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Change-Id: I2047f935bdffbc72b066d905b0766bfb1f3cc57a"},{"line_number":71,"context_line":"Signed-off-by: Kashyap Chamarthy \u003ckchamart@redhat.com\u003e"},{"line_number":72,"context_line":"Thanks: Sean Mooney \u003cwork@seanmooney.info\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3fa7e38b_cc5e5543","line":69,"updated":"2019-09-24 16:27:17.000000000","message":"we proably should file a bug for this so we can backport","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":6962,"name":"Kashyap Chamarthy","email":"kchamart@redhat.com","username":"kashyapc"},"change_message_id":"42b84bcb02460d00ac0fd7677298622dabde5519","unresolved":false,"context_lines":[{"line_number":66,"context_line":""},{"line_number":67,"context_line":"[*] https://bugzilla.redhat.com/show_bug.cgi?id\u003d1010885#c2 --"},{"line_number":68,"context_line":"    kvm_init_vcpu failed: Cannot allocate memory in NUMA"},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Change-Id: I2047f935bdffbc72b066d905b0766bfb1f3cc57a"},{"line_number":71,"context_line":"Signed-off-by: Kashyap Chamarthy \u003ckchamart@redhat.com\u003e"},{"line_number":72,"context_line":"Thanks: Sean Mooney \u003cwork@seanmooney.info\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3fa7e38b_af42eb3c","line":69,"in_reply_to":"3fa7e38b_cc5e5543","updated":"2019-09-24 16:37:37.000000000","message":"Yeah, that was also on my TODO; but didn\u0027t mention; will do.","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"333535c0f206514aa21b1602871e94a9e8a56ed4","unresolved":false,"context_lines":[{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Change-Id: I2047f935bdffbc72b066d905b0766bfb1f3cc57a"},{"line_number":71,"context_line":"Signed-off-by: Kashyap Chamarthy \u003ckchamart@redhat.com\u003e"},{"line_number":72,"context_line":"Thanks: Sean Mooney \u003cwork@seanmooney.info\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3fa7e38b_8c54dd22","line":72,"updated":"2019-09-24 16:27:17.000000000","message":"im not sure this is required but ok","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":6962,"name":"Kashyap Chamarthy","email":"kchamart@redhat.com","username":"kashyapc"},"change_message_id":"42b84bcb02460d00ac0fd7677298622dabde5519","unresolved":false,"context_lines":[{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Change-Id: I2047f935bdffbc72b066d905b0766bfb1f3cc57a"},{"line_number":71,"context_line":"Signed-off-by: Kashyap Chamarthy \u003ckchamart@redhat.com\u003e"},{"line_number":72,"context_line":"Thanks: Sean Mooney \u003cwork@seanmooney.info\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"3fa7e38b_4f7d777d","line":72,"in_reply_to":"3fa7e38b_8c54dd22","updated":"2019-09-24 16:37:37.000000000","message":"Was trying to to acknowledge your time, as you were very responsive in answering my questions and reviewing my assumptions.","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"}],"nova/virt/libvirt/config.py":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"333535c0f206514aa21b1602871e94a9e8a56ed4","unresolved":false,"context_lines":[{"line_number":2363,"context_line":"        super(LibvirtConfigGuestNUMATuneMemory, self).__init__("},{"line_number":2364,"context_line":"            root_name\u003d\"memory\", **kwargs)"},{"line_number":2365,"context_line":""},{"line_number":2366,"context_line":"        self.mode \u003d \"preferred\""},{"line_number":2367,"context_line":"        self.nodeset \u003d []"},{"line_number":2368,"context_line":""},{"line_number":2369,"context_line":"    def format_dom(self):"}],"source_content_type":"text/x-python","patch_set":1,"id":"3fa7e38b_09445f1a","line":2366,"updated":"2019-09-24 16:27:17.000000000","message":"we might want to add a code comment here to state why this is done.","commit_id":"e471ad57b54188f66eee39412baf581aa9f22fca"},{"author":{"_account_id":6962,"name":"Kashyap Chamarthy","email":"kchamart@redhat.com","username":"kashyapc"},"change_message_id":"42b84bcb02460d00ac0fd7677298622dabde5519","unresolved":false,"context_lines":[{"line_number":2363,"context_line":"        super(LibvirtConfigGuestNUMATuneMemory, self).__init__("},{"line_number":2364,"context_line":"            root_name\u003d\"memory\", **kwargs)"},{"line_number":2365,"context_line":""},{"line_number":2366,"context_line":"        self.mode \u003d \"preferred\""},{"line_number":2367,"context_line":"        self.nodeset \u003d []"},{"line_number":2368,"context_line":""},{"line_number":2369,"context_line":"    def format_dom(self):"}],"source_content_type":"text/x-python","patch_set":1,"id":"3fa7e38b_4fa257ff","line":2366,"in_reply_to":"3fa7e38b_09445f1a","updated":"2019-09-24 16:37:37.000000000","message":"Yes, can do.","commit_id":"e471ad57b54188f66eee39412baf581aa9f22fca"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"333535c0f206514aa21b1602871e94a9e8a56ed4","unresolved":false,"context_lines":[{"line_number":2363,"context_line":"        super(LibvirtConfigGuestNUMATuneMemory, self).__init__("},{"line_number":2364,"context_line":"            root_name\u003d\"memory\", **kwargs)"},{"line_number":2365,"context_line":""},{"line_number":2366,"context_line":"        self.mode \u003d \"preferred\""},{"line_number":2367,"context_line":"        self.nodeset \u003d []"},{"line_number":2368,"context_line":""},{"line_number":2369,"context_line":"    def format_dom(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3fa7e38b_2c4e0997","line":2366,"range":{"start_line":2366,"start_character":6,"end_line":2366,"end_character":31},"updated":"2019-09-24 16:27:17.000000000","message":"we might want to add a code comment here to state why this is set to preferred.","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"0c56ee1b2c98a02059bd45d6056b8c834f274a3c","unresolved":false,"context_lines":[{"line_number":2379,"context_line":"        # allocation fails and in turn the VM boot fails with:"},{"line_number":2380,"context_line":"        # \"kvm_init_vcpu failed: Cannot allocate memory\"."},{"line_number":2381,"context_line":"        #"},{"line_number":2382,"context_line":"        # This kernel commit ee6268ba3a68 (\"KVM: x86: Skip pae_root"},{"line_number":2383,"context_line":"        # shadow allocation if tdp enabled\", 2018-07-25) should fix the"},{"line_number":2384,"context_line":"        # \"can\u0027t boot a VM because of missing \"DMA32\" zone\" problem)."},{"line_number":2385,"context_line":"        # It is available in upstream Linux 4.19 onwards (or backported"},{"line_number":2386,"context_line":"        # to older kernels by Linux vendors; and the fix originated from"},{"line_number":2387,"context_line":"        # the same memory allocation problem discussed here:"},{"line_number":2388,"context_line":"        # https://lkml.org/lkml/2018/7/24/843"},{"line_number":2389,"context_line":"        #"},{"line_number":2390,"context_line":"        # From Nova point of view, relax the above \"DMA32\" zone"},{"line_number":2391,"context_line":"        # requirement, by using the \u0027preferred\u0027 mode, which means: turn"}],"source_content_type":"text/x-python","patch_set":3,"id":"3fa7e38b_c258cafb","line":2388,"range":{"start_line":2382,"start_character":0,"end_line":2388,"end_character":45},"updated":"2019-10-03 09:14:04.000000000","message":"This fixed things, but what broke it in the first place? It seems we\u0027ve had an uptick in these issues recently, based on openstack-discuss posts let alone BZs, so I\u0027m guessing this was introduced by a change at some point in the past?","commit_id":"e74ed41489c2772fe1845edaca2e29f1ddb10d77"},{"author":{"_account_id":6962,"name":"Kashyap Chamarthy","email":"kchamart@redhat.com","username":"kashyapc"},"change_message_id":"39208b36f8583bc371c726bc2917e3fa43647352","unresolved":false,"context_lines":[{"line_number":2379,"context_line":"        # allocation fails and in turn the VM boot fails with:"},{"line_number":2380,"context_line":"        # \"kvm_init_vcpu failed: Cannot allocate memory\"."},{"line_number":2381,"context_line":"        #"},{"line_number":2382,"context_line":"        # This kernel commit ee6268ba3a68 (\"KVM: x86: Skip pae_root"},{"line_number":2383,"context_line":"        # shadow allocation if tdp enabled\", 2018-07-25) should fix the"},{"line_number":2384,"context_line":"        # \"can\u0027t boot a VM because of missing \"DMA32\" zone\" problem)."},{"line_number":2385,"context_line":"        # It is available in upstream Linux 4.19 onwards (or backported"},{"line_number":2386,"context_line":"        # to older kernels by Linux vendors; and the fix originated from"},{"line_number":2387,"context_line":"        # the same memory allocation problem discussed here:"},{"line_number":2388,"context_line":"        # https://lkml.org/lkml/2018/7/24/843"},{"line_number":2389,"context_line":"        #"},{"line_number":2390,"context_line":"        # From Nova point of view, relax the above \"DMA32\" zone"},{"line_number":2391,"context_line":"        # requirement, by using the \u0027preferred\u0027 mode, which means: turn"}],"source_content_type":"text/x-python","patch_set":3,"id":"3fa7e38b_27d0a9f4","line":2388,"range":{"start_line":2382,"start_character":0,"end_line":2388,"end_character":45},"in_reply_to":"3fa7e38b_c258cafb","updated":"2020-01-16 15:33:01.000000000","message":"Reading the 2-line kernel commit, it was a latent issue:\n\nhttps://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id\u003dee6268ba3a68","commit_id":"e74ed41489c2772fe1845edaca2e29f1ddb10d77"}],"releasenotes/notes/Memory-allocation-mode-changes-to-strict-f97b3c13be646b5c.yaml":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"333535c0f206514aa21b1602871e94a9e8a56ed4","unresolved":false,"context_lines":[{"line_number":5,"context_line":"    Nova used to default the memory allocation mode to ``strict``.  The"},{"line_number":6,"context_line":"    ``strict`` mode mandates that the NUMA node have a \"DMA32\" region"},{"line_number":7,"context_line":"    (it is one of the \"zones\" into which the Linux kernel divides the"},{"line_number":8,"context_line":"    system memory).  On a NUMA node without the \"DMA32\" region,"},{"line_number":9,"context_line":"    instances will fail to launch.  To fix this, the default memory"},{"line_number":10,"context_line":"    allocation mode is now changed to ``preferred``, so that this"},{"line_number":11,"context_line":"    instances can now be launched on NUMA nodes _without_ a \"DMA32\""}],"source_content_type":"text/x-yaml","patch_set":2,"id":"3fa7e38b_acb3d997","line":8,"range":{"start_line":8,"start_character":26,"end_line":8,"end_character":55},"updated":"2019-09-24 16:27:17.000000000","message":"normally we see a DMA32 region per numa node but oh some hardware we only have 1 DMA32 region. i have n\n\ni belive this is configurable in some uefi/bios configuration in the secation wehre you modify above 4G decodeing but i dont think we can rely on that. i cannot find a way to change the behavior on any of the systems i have acesss too and they all have a DMA32 region per numa node","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":6962,"name":"Kashyap Chamarthy","email":"kchamart@redhat.com","username":"kashyapc"},"change_message_id":"42b84bcb02460d00ac0fd7677298622dabde5519","unresolved":false,"context_lines":[{"line_number":5,"context_line":"    Nova used to default the memory allocation mode to ``strict``.  The"},{"line_number":6,"context_line":"    ``strict`` mode mandates that the NUMA node have a \"DMA32\" region"},{"line_number":7,"context_line":"    (it is one of the \"zones\" into which the Linux kernel divides the"},{"line_number":8,"context_line":"    system memory).  On a NUMA node without the \"DMA32\" region,"},{"line_number":9,"context_line":"    instances will fail to launch.  To fix this, the default memory"},{"line_number":10,"context_line":"    allocation mode is now changed to ``preferred``, so that this"},{"line_number":11,"context_line":"    instances can now be launched on NUMA nodes _without_ a \"DMA32\""}],"source_content_type":"text/x-yaml","patch_set":2,"id":"3fa7e38b_cfade7e9","line":8,"range":{"start_line":8,"start_character":26,"end_line":8,"end_character":55},"in_reply_to":"3fa7e38b_acb3d997","updated":"2019-09-24 16:37:37.000000000","message":"\u003e normally we see a DMA32 region per numa node but oh some hardware\n \u003e we only have 1 DMA32 region. i have n\n\nIs that an unfinished sentence?  (Please rephrase the last part.)\n\n \u003e \n \u003e i belive this is configurable in some uefi/bios configuration in\n \u003e the secation wehre you modify above 4G decodeing but i dont think\n \u003e we can rely on that. i cannot find a way to change the behavior on\n \u003e any of the systems i have acesss too and they all have a DMA32\n \u003e region per numa node\n\nNod; I didn\u0027t wanted to get into that detail here.","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"39253389d747785ea65e14fb5c0f588c5337a2f2","unresolved":false,"context_lines":[{"line_number":5,"context_line":"    Nova used to default the memory allocation mode to ``strict``.  The"},{"line_number":6,"context_line":"    ``strict`` mode mandates that the NUMA node have a \"DMA32\" region"},{"line_number":7,"context_line":"    (it is one of the \"zones\" into which the Linux kernel divides the"},{"line_number":8,"context_line":"    system memory).  On a NUMA node without the \"DMA32\" region,"},{"line_number":9,"context_line":"    instances will fail to launch.  To fix this, the default memory"},{"line_number":10,"context_line":"    allocation mode is now changed to ``preferred``, so that this"},{"line_number":11,"context_line":"    instances can now be launched on NUMA nodes _without_ a \"DMA32\""}],"source_content_type":"text/x-yaml","patch_set":2,"id":"3fa7e38b_ef25435d","line":8,"range":{"start_line":8,"start_character":26,"end_line":8,"end_character":55},"in_reply_to":"3fa7e38b_cfade7e9","updated":"2019-09-24 16:48:11.000000000","message":"this comment was more for reviewers. i dont expect you to update this. it is fine as it is.\n\nand ya the unfinished sentance was \"i have not seen this on any hardware i have worked with\" but then i decied to test the hardware i have at home to confirm.","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"},{"author":{"_account_id":6962,"name":"Kashyap Chamarthy","email":"kchamart@redhat.com","username":"kashyapc"},"change_message_id":"d9bd7c198d3f57614e0d1f49262606a5b60b67ed","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":2,"id":"3fa7e38b_a9a42ba7","line":13,"updated":"2019-09-24 15:36:24.000000000","message":"Note to self:\n\nAlso add an \"upgrade:\" note here, something like:\n\nWhen migrating an instance from a Compute node with \u0027strict\u0027 to one with \u0027preferred\u0027 mode, make sure to hard-reboot the instances for the new mode (\u0027preferred\u0027) to take effect.","commit_id":"a8c482fed2986f8a54d8823639cb139419d18f43"}]}