Beispiel #1
0
def raw_data(request, crash_id, extension, name=None):
    api = models.RawCrash()
    if extension == "json":
        format = "meta"
        content_type = "application/json"
    elif extension == "dmp":
        format = "raw"
        content_type = "application/octet-stream"
    elif extension == "json.gz" and name == "memory_report":
        # Note, if the name is 'memory_report' it will fetch a raw
        # crash with name and the files in the memory_report bucket
        # are already gzipped.
        # This is important because it means we don't need to gzip
        # the HttpResponse below.
        format = "raw"
        content_type = "application/octet-stream"
    else:
        raise NotImplementedError(extension)

    data = api.get(crash_id=crash_id, format=format, name=name)
    response = http.HttpResponse(content_type=content_type)
    if extension == "json":
        response.write(json.dumps(data))
    else:
        response.write(data)
    return response
Beispiel #2
0
def crash_verify(request):
    """Verifies crash data in crash data destinations"""
    crash_id = request.GET.get('crash_id', None)

    if not crash_id or not is_crash_id_valid(crash_id):
        return http.JsonResponse({'error': 'unknown crash id'}, status=400)

    data = {'uuid': crash_id}

    # Check S3 crash bucket for raw and processed crash data
    raw_api = models.RawCrash()
    try:
        raw_api.get(crash_id=crash_id, dont_cache=True, refresh_cache=True)
        has_raw_crash = True
    except CrashIDNotFound:
        has_raw_crash = False
    data['s3_raw_crash'] = has_raw_crash

    processed_api = models.ProcessedCrash()
    try:
        processed_api.get(crash_id=crash_id,
                          dont_cache=True,
                          refresh_cache=True)
        has_processed_crash = True
    except CrashIDNotFound:
        has_processed_crash = False
    data['s3_processed_crash'] = has_processed_crash

    # Check Elasticsearch for crash data
    supersearch_api = supersearch_models.SuperSearch()
    params = {
        '_columns': ['uuid'],
        '_results_number': 1,
        'uuid': crash_id,
        'dont_cache': True,
        'refresh_cache': True
    }
    results = supersearch_api.get(**params)
    data['elasticsearch_crash'] = (results['total'] == 1
                                   and results['hits'][0]['uuid'] == crash_id)

    # Check S3 telemetry bucket for crash data
    telemetry_api = models.TelemetryCrash()
    try:
        telemetry_api.get(crash_id=crash_id,
                          dont_cache=True,
                          refresh_cache=True)
        has_telemetry_crash = True
    except CrashIDNotFound:
        has_telemetry_crash = False
    data['s3_telemetry_crash'] = has_telemetry_crash

    # NOTE(willkg): This doesn't check postgres because that's being phased
    # out.

    return http.HttpResponse(json.dumps(data), content_type='application/json')
Beispiel #3
0
def crash_verify(request):
    """Verifies crash data in crash data destinations"""
    crash_id = request.GET.get("crash_id", None)

    if not crash_id or not is_crash_id_valid(crash_id):
        return http.JsonResponse({"error": "unknown crash id"}, status=400)

    data = {"uuid": crash_id}

    # Check S3 crash bucket for raw and processed crash data
    raw_api = models.RawCrash()
    try:
        raw_api.get(crash_id=crash_id, dont_cache=True, refresh_cache=True)
        has_raw_crash = True
    except CrashIDNotFound:
        has_raw_crash = False
    data["s3_raw_crash"] = has_raw_crash

    processed_api = models.ProcessedCrash()
    try:
        processed_api.get(crash_id=crash_id,
                          dont_cache=True,
                          refresh_cache=True)
        has_processed_crash = True
    except CrashIDNotFound:
        has_processed_crash = False
    data["s3_processed_crash"] = has_processed_crash

    # Check Elasticsearch for crash data
    supersearch_api = supersearch_models.SuperSearch()
    params = {
        "_columns": ["uuid"],
        "_results_number": 1,
        "uuid": crash_id,
        "dont_cache": True,
        "refresh_cache": True,
    }
    results = supersearch_api.get(**params)
    data["elasticsearch_crash"] = (results["total"] == 1
                                   and results["hits"][0]["uuid"] == crash_id)

    # Check S3 telemetry bucket for crash data
    telemetry_api = models.TelemetryCrash()
    try:
        telemetry_api.get(crash_id=crash_id,
                          dont_cache=True,
                          refresh_cache=True)
        has_telemetry_crash = True
    except CrashIDNotFound:
        has_telemetry_crash = False
    data["s3_telemetry_crash"] = has_telemetry_crash

    return http.HttpResponse(json.dumps(data), content_type="application/json")
Beispiel #4
0
def report_index(request, crash_id, default_context=None):
    valid_crash_id = utils.find_crash_id(crash_id)
    if not valid_crash_id:
        return http.HttpResponseBadRequest("Invalid crash ID")

    # Sometimes, in Socorro we use a prefix on the crash ID. Usually it's
    # 'bp-' but this is configurable.
    # If you try to use this to reach the perma link for a crash, it should
    # redirect to the report index with the correct crash ID.
    if valid_crash_id != crash_id:
        return redirect(
            reverse("crashstats:report_index", args=(valid_crash_id, )))

    context = default_context or {}
    context["crash_id"] = crash_id

    refresh_cache = request.GET.get("refresh") == "cache"

    raw_api = models.RawCrash()
    try:
        context["raw"] = raw_api.get(crash_id=crash_id,
                                     refresh_cache=refresh_cache)
    except CrashIDNotFound:
        # If the raw crash can't be found, we can't do much.
        return render(request,
                      "crashstats/report_index_not_found.html",
                      context,
                      status=404)
    utils.enhance_raw(context["raw"])

    context["your_crash"] = (request.user.is_active and
                             context["raw"].get("Email") == request.user.email)

    api = models.UnredactedCrash()
    try:
        context["report"] = api.get(crash_id=crash_id,
                                    refresh_cache=refresh_cache)
    except CrashIDNotFound:
        # ...if we haven't already done so.
        cache_key = "priority_job:{}".format(crash_id)
        if not cache.get(cache_key):
            priority_api = models.PriorityJob()
            priority_api.post(crash_ids=[crash_id])
            cache.set(cache_key, True, 60)
        return render(request, "crashstats/report_index_pending.html", context)

    if "json_dump" in context["report"]:
        json_dump = context["report"]["json_dump"]
        if "sensitive" in json_dump and not request.user.has_perm(
                "crashstats.view_pii"):
            del json_dump["sensitive"]
        context["raw_stackwalker_output"] = json.dumps(json_dump,
                                                       sort_keys=True,
                                                       indent=4,
                                                       separators=(",", ": "))
        utils.enhance_json_dump(json_dump, settings.VCS_MAPPINGS)
        parsed_dump = json_dump
    else:
        context["raw_stackwalker_output"] = "No dump available"
        parsed_dump = {}

    # NOTE(willkg): pull cpu count from parsed_dump if it's not in report;
    # remove in July 2019
    if "cpu_count" not in context["report"]:
        context["report"]["cpu_count"] = parsed_dump.get("system_info",
                                                         {}).get("cpu_count")
    # NOTE(willkg): "cpu_name" is deprecated, but populate "cpu_arch" if
    # cpu_arch is empty; remove in July 2019.
    if "cpu_arch" not in context["report"]:
        context["report"]["cpu_arch"] = context["report"]["cpu_name"]

    context["crashing_thread"] = parsed_dump.get("crash_info",
                                                 {}).get("crashing_thread")
    if context["report"]["signature"].startswith("shutdownhang"):
        # For shutdownhang signatures, we want to use thread 0 as the
        # crashing thread, because that's the thread that actually contains
        # the useful data about what happened.
        context["crashing_thread"] = 0

    context["parsed_dump"] = parsed_dump
    context["bug_product_map"] = settings.BUG_PRODUCT_MAP

    context["bug_associations"] = list(
        models.BugAssociation.objects.filter(
            signature=context["report"]["signature"]).values(
                "bug_id", "signature").order_by("-bug_id"))

    context["raw_keys"] = []
    if request.user.has_perm("crashstats.view_pii"):
        # hold nothing back
        context["raw_keys"] = context["raw"].keys()
    else:
        context["raw_keys"] = [
            x for x in context["raw"] if x in models.RawCrash.API_ALLOWLIST()
        ]
    # Sort keys case-insensitively
    context["raw_keys"] = sorted(context["raw_keys"], key=lambda s: s.lower())

    if request.user.has_perm("crashstats.view_rawdump"):
        context["raw_dump_urls"] = [
            reverse("crashstats:raw_data", args=(crash_id, "dmp")),
            reverse("crashstats:raw_data", args=(crash_id, "json")),
        ]
        if context["raw"].get("additional_minidumps"):
            suffixes = [
                x.strip()
                for x in context["raw"]["additional_minidumps"].split(",")
                if x.strip()
            ]
            for suffix in suffixes:
                name = "upload_file_minidump_%s" % (suffix, )
                context["raw_dump_urls"].append(
                    reverse("crashstats:raw_data_named",
                            args=(crash_id, name, "dmp")))
        if (context["raw"].get("ContainsMemoryReport")
                and context["report"].get("memory_report")
                and not context["report"].get("memory_report_error")):
            context["raw_dump_urls"].append(
                reverse(
                    "crashstats:raw_data_named",
                    args=(crash_id, "memory_report", "json.gz"),
                ))

    # Add descriptions to all fields.
    all_fields = SuperSearchFields().get()
    descriptions = {}
    for field in all_fields.values():
        key = "{}.{}".format(field["namespace"], field["in_database_name"])
        descriptions[key] = "{} Search: {}".format(
            field.get("description", "").strip()
            or "No description for this field.",
            field["is_exposed"] and field["name"] or "N/A",
        )

    def make_raw_crash_key(key):
        """In the report_index.html template we need to create a key
        that we can use to look up against the 'fields_desc' dict.
        Because you can't do something like this in jinja::

            {{ fields_desc.get(u'raw_crash.{}'.format(key), empty_desc) }}

        we do it here in the function instead.
        The trick is that the lookup key has to be a unicode object or
        else you get UnicodeEncodeErrors in the template rendering.
        """
        return "raw_crash.{}".format(key)

    context["make_raw_crash_key"] = make_raw_crash_key
    context["fields_desc"] = descriptions
    context["empty_desc"] = "No description for this field. Search: unknown"

    context["BUG_PRODUCT_MAP"] = settings.BUG_PRODUCT_MAP

    # report.addons used to be a list of lists.
    # In https://bugzilla.mozilla.org/show_bug.cgi?id=1250132
    # we changed it from a list of lists to a list of strings, using
    # a ':' to split the name and version.
    # See https://bugzilla.mozilla.org/show_bug.cgi?id=1250132#c7
    # Considering legacy, let's tackle both.
    # In late 2017, this code is going to be useless and can be removed.
    if context["report"].get("addons") and isinstance(
            context["report"]["addons"][0], (list, tuple)):
        # This is the old legacy format. This crash hasn't been processed
        # the new way.
        context["report"]["addons"] = [
            ":".join(x) for x in context["report"]["addons"]
        ]

    content = loader.render_to_string("crashstats/report_index.html", context,
                                      request)
    utf8_content = content.encode("utf-8", errors="backslashreplace")
    return HttpResponse(utf8_content, charset="utf-8")
Beispiel #5
0
def report_index(request, crash_id, default_context=None):
    valid_crash_id = utils.find_crash_id(crash_id)
    if not valid_crash_id:
        return http.HttpResponseBadRequest('Invalid crash ID')

    # Sometimes, in Socorro we use a prefix on the crash ID. Usually it's
    # 'bp-' but this is configurable.
    # If you try to use this to reach the perma link for a crash, it should
    # redirect to the report index with the correct crash ID.
    if valid_crash_id != crash_id:
        return redirect(
            reverse('crashstats:report_index', args=(valid_crash_id, )))

    context = default_context or {}
    context['crash_id'] = crash_id

    refresh_cache = request.GET.get('refresh') == 'cache'

    raw_api = models.RawCrash()
    try:
        context['raw'] = raw_api.get(crash_id=crash_id,
                                     refresh_cache=refresh_cache)
    except CrashIDNotFound:
        # If the raw crash can't be found, we can't do much.
        return render(request,
                      'crashstats/report_index_not_found.html',
                      context,
                      status=404)
    utils.enhance_raw(context['raw'])

    context['your_crash'] = (request.user.is_active and
                             context['raw'].get('Email') == request.user.email)

    api = models.UnredactedCrash()
    try:
        context['report'] = api.get(crash_id=crash_id,
                                    refresh_cache=refresh_cache)
    except CrashIDNotFound:
        # ...if we haven't already done so.
        cache_key = 'priority_job:{}'.format(crash_id)
        if not cache.get(cache_key):
            priority_api = models.Priorityjob()
            priority_api.post(crash_ids=[crash_id])
            cache.set(cache_key, True, 60)
        return render(request, 'crashstats/report_index_pending.html', context)

    if 'json_dump' in context['report']:
        json_dump = context['report']['json_dump']
        if 'sensitive' in json_dump and not request.user.has_perm(
                'crashstats.view_pii'):
            del json_dump['sensitive']
        context['raw_stackwalker_output'] = json.dumps(json_dump,
                                                       sort_keys=True,
                                                       indent=4,
                                                       separators=(',', ': '))
        utils.enhance_json_dump(json_dump, settings.VCS_MAPPINGS)
        parsed_dump = json_dump
    else:
        context['raw_stackwalker_output'] = 'No dump available'
        parsed_dump = {}

    # NOTE(willkg): pull cpu count from parsed_dump if it's not in report;
    # remove in July 2019
    if 'cpu_count' not in context['report']:
        context['report']['cpu_count'] = parsed_dump.get('system_info',
                                                         {}).get('cpu_count')
    # NOTE(willkg): "cpu_name" is deprecated, but populate "cpu_arch" if
    # cpu_arch is empty; remove in July 2019.
    if 'cpu_arch' not in context['report']:
        context['report']['cpu_arch'] = context['report']['cpu_name']

    context['crashing_thread'] = parsed_dump.get('crash_info',
                                                 {}).get('crashing_thread')
    if context['report']['signature'].startswith('shutdownhang'):
        # For shutdownhang signatures, we want to use thread 0 as the
        # crashing thread, because that's the thread that actually contains
        # the useful data about what happened.
        context['crashing_thread'] = 0

    context['parsed_dump'] = parsed_dump
    context['bug_product_map'] = settings.BUG_PRODUCT_MAP

    context['bug_associations'] = list(
        models.BugAssociation.objects.filter(
            signature=context['report']['signature']).values(
                'bug_id', 'signature').order_by('-bug_id'))

    context['raw_keys'] = []
    if request.user.has_perm('crashstats.view_pii'):
        # hold nothing back
        context['raw_keys'] = context['raw'].keys()
    else:
        context['raw_keys'] = [
            x for x in context['raw'] if x in models.RawCrash.API_WHITELIST()
        ]
    # Sort keys case-insensitively
    context['raw_keys'] = sorted(context['raw_keys'], key=lambda s: s.lower())

    if request.user.has_perm('crashstats.view_rawdump'):
        context['raw_dump_urls'] = [
            reverse('crashstats:raw_data', args=(crash_id, 'dmp')),
            reverse('crashstats:raw_data', args=(crash_id, 'json'))
        ]
        if context['raw'].get('additional_minidumps'):
            suffixes = [
                x.strip()
                for x in context['raw']['additional_minidumps'].split(',')
                if x.strip()
            ]
            for suffix in suffixes:
                name = 'upload_file_minidump_%s' % (suffix, )
                context['raw_dump_urls'].append(
                    reverse('crashstats:raw_data_named',
                            args=(crash_id, name, 'dmp')))
        if (context['raw'].get('ContainsMemoryReport')
                and context['report'].get('memory_report')
                and not context['report'].get('memory_report_error')):
            context['raw_dump_urls'].append(
                reverse('crashstats:raw_data_named',
                        args=(crash_id, 'memory_report', 'json.gz')))

    # Add descriptions to all fields.
    all_fields = SuperSearchFields().get()
    descriptions = {}
    for field in all_fields.values():
        key = '{}.{}'.format(field['namespace'], field['in_database_name'])
        descriptions[key] = '{} Search: {}'.format(
            field.get('description', '').strip()
            or 'No description for this field.',
            field['is_exposed'] and field['name'] or 'N/A',
        )

    def make_raw_crash_key(key):
        """In the report_index.html template we need to create a key
        that we can use to look up against the 'fields_desc' dict.
        Because you can't do something like this in jinja::

            {{ fields_desc.get(u'raw_crash.{}'.format(key), empty_desc) }}

        we do it here in the function instead.
        The trick is that the lookup key has to be a unicode object or
        else you get UnicodeEncodeErrors in the template rendering.
        """
        return u'raw_crash.{}'.format(key)

    context['make_raw_crash_key'] = make_raw_crash_key
    context['fields_desc'] = descriptions
    context['empty_desc'] = 'No description for this field. Search: unknown'

    context['BUG_PRODUCT_MAP'] = settings.BUG_PRODUCT_MAP

    # report.addons used to be a list of lists.
    # In https://bugzilla.mozilla.org/show_bug.cgi?id=1250132
    # we changed it from a list of lists to a list of strings, using
    # a ':' to split the name and version.
    # See https://bugzilla.mozilla.org/show_bug.cgi?id=1250132#c7
    # Considering legacy, let's tackle both.
    # In late 2017, this code is going to be useless and can be removed.
    if (context['report'].get('addons')
            and isinstance(context['report']['addons'][0], (list, tuple))):
        # This is the old legacy format. This crash hasn't been processed
        # the new way.
        context['report']['addons'] = [
            ':'.join(x) for x in context['report']['addons']
        ]

    return render(request, 'crashstats/report_index.html', context)