EVOLUTION-NINJA

Edit File: cassandra.py

import agent_util
import csv
import os
import sys

if sys.version[0] == "3":
    from io import StringIO
else:
    from StringIO import StringIO
from agent_util import float

textkey_mapping = {}

def get_cmd_location(config):
    program_location = ""
    if "cassandra_installed_location" in config:
        cassandra_installed_location = config["cassandra_installed_location"]
        program_location = os.path.join(cassandra_installed_location, "bin")
        cmd = os.path.join(program_location, "nodetool")
        if not os.path.isfile(cmd):
            cmd = agent_util.which("nodetool")
    else:
        cmd = agent_util.which("nodetool", exc=True)
    return cmd

def name_to_textkey(name):
    textkey = name.strip()
    textkey = textkey.replace(" ", "_")
    textkey = textkey.replace("(", "").replace(")", "")
    textkey_mapping[textkey] = name
    return textkey

def get_cfstats(config, keyspace=None, table=None):
    cmd = get_cmd_location(config)
    cmd += " cfstats"
    if keyspace and table:
        cmd += " %s.%s" % (keyspace, table)

status, output = agent_util.execute_command(cmd)
    if status != 0:
        raise Exception(output)

lines = output.split("\n")
    data = {}

current_keyspace = None
    current_table = None

for line in lines:
        if line.startswith("Keyspace:"):
            current_keyspace = line.strip().split()[-1]
            current_table = None
            data[current_keyspace] = {}
            data[current_keyspace]["tables"] = {}
            continue

if line.strip().startswith("Table:"):
            current_table = line.strip().split()[-1]
            data[current_keyspace]["tables"][current_table] = {}
            continue

# Skip empty lines and "-------" divider lines
        if not line.strip():
            continue
        if not line.replace("-", "").strip():
            continue

# Get the key/value pair on the line, get rid of "ms" and convert to a float.  Use
        # try/except to convert NaN to None
        fields = line.strip().split(":")
        key = fields[0]
        value = fields[1]
        key = name_to_textkey(key)
        value = value.strip()
        value = value.replace("ms", "").strip()
        try:
            if value.lower() == "nan":
                raise Exception()
            value = float(value)
        except:
            value = None

if current_table:
            data[current_keyspace]["tables"][current_table][key] = value
        else:
            data[current_keyspace][key] = value

return data

def get_cfhistograms(config, keyspace, table):
    cmd = get_cmd_location(config)
    cmd += " cfhistograms -- %s %s" % (keyspace, table)
    status, output = agent_util.execute_command(cmd)
    if status != 0:
        raise Exception(output)

lines = output.split("\n")

output = agent_util.StringIO(output)
    parsed_output = list(csv.reader(output, delimiter="\t"))
    indices = [i for i, x in enumerate(parsed_output) if x == []]

cfhistograms_dict = dict(
        (parsed_output[i - 2][0], parsed_output[i - 1][0])
        for i in indices
        if i - 2 >= 0
    )

return cfhistograms_dict

def get_tpstats(config):
    cmd = get_cmd_location(config)
    status, output = agent_util.execute_command(cmd + " tpstats")
    if status != 0:
        raise Exception(output)
    output = agent_util.StringIO(output)
    parsed_output = list(csv.reader(output, delimiter="\t"))
    header_line = [x.strip() for x in parsed_output[0][0].split("  ") if x]
    pool_tasks = []
    dropped_message = []
    header = []
    for values in parsed_output[1:]:
        if values:
            v = [x.strip() for x in values[0].split("  ") if x]
            if len(v) == 2:
                if not header:
                    header = [x.strip() for x in values[0].split("  ") if x]
                else:
                    dropped_message.append(dict(zip(header, v)))
            else:
                pool_tasks.append(dict(zip(header_line, v)))

return pool_tasks, dropped_message

class CassandraPlugin(agent_util.Plugin):
    textkey = "cassandra"
    label = "Cassandra"

@classmethod
    def get_metadata(self, config):
        status = agent_util.SUPPORTED
        msg = None
        program_location = ""
        if "cassandra_installed_location" in config:
            cassandra_installed_location = config["cassandra_installed_location"]
            program_location = os.path.join(cassandra_installed_location, "bin")
        # check if cassandra is even installed
        installed = agent_util.which("cassandra")
        if not installed:
            self.log.info("cassandra binary not found")
            status = agent_util.UNSUPPORTED
            msg = "cassandra binary not found"
            return {}

tables = []
        keyspaces = []
        pool_tasks = []
        dropped_message = []
        if status is agent_util.SUPPORTED:
            try:
                cfstats = get_cfstats(config)
                keyspaces = cfstats.keys()
                for keyspace in cfstats:
                    for table in cfstats[keyspace]["tables"]:
                        tables.append(keyspace + "." + table)

pool_tasks, dropped_message = get_tpstats(config)
                pool_tasks = [p["Pool Name"] for p in pool_tasks]
                dropped_message = [m["Message type"] for m in dropped_message]

except:
                status = agent_util.MISCONFIGURED
                self.log.error("couldn't get cassandra stats")
                msg = "Couldn't get Cassandra stats, make sure cassandra is installed and cassandra configuration file is valid."

tables.sort()
        keyspaces.sort()

metadata = {}

# Add keyspace-level metrics
        keyspace = keyspaces[0]
        for key in cfstats[keyspace]:
            if key != "tables":
                metadata[key] = {
                    "label": textkey_mapping[key],
                    "options": keyspaces,
                    "status": status,
                    "error_message": msg,
                }

# Add table-level metrics
        table = cfstats[keyspace]["tables"].keys()[0]
        for key in cfstats[keyspace]["tables"][table].keys():
            metadata[key] = {
                "label": textkey_mapping[key],
                "options": tables,
                "status": status,
                "error_message": msg,
            }

"""
        metadata = {
            "thread_pool_task.active": {
                "label": "Thread pool task: active",
                "options": pool_tasks,
                "status": status,
                "error_message": msg
            },
            "thread_pool_task.completed": {
                "label": "Thread pool task: completed",
                "options": pool_tasks,
                "status": status,
                "error_message": msg
            },
            "thread_pool_task.blocked": {
                "label": "Thread pool task: blocked",
                "options": pool_tasks,
                "status": status,
                "error_message": msg
            },
            "thread_pool_task.pending": {
                "label": "Thread pool task: pending",
                "options": pool_tasks,
                "status": status,
                "error_message": msg
            },
            "dropped_messages": {
                "label": "Dropped Messages",
                "options": dropped_message,
                "status": status,
                "error_message": msg
            },

}
        """
        return metadata

def check(self, textkey, data, config):
        res = 0
        if data:
            table = ""
            if len(data.split(".")) == 2:
                keyspace, table = data.split(".")
            else:
                keyspace = data

if textkey in ["column_count", "row_size", "row_count"]:
                cfhistograms_dict = get_cfhistograms(config, keyspace, table)
                res = cfhistograms_dict[mapping[textkey]]
                if textkey == "column_count":
                    try:
                        res = res.split(" cells")[0]
                    except:
                        res = 0
                else:
                    if textkey == "row_size":
                        try:
                            res = res.split(" bytes: ")[0]
                        except:
                            res = 0
                    else:
                        try:
                            res = res.split(" bytes: ")[1]
                        except:
                            res = 0

elif "thread_pool_task" in textkey or textkey == "dropped_messages":
                pool_tasks, dropped_message = get_tpstats(config)
                if "thread_pool_task" in textkey:
                    for p in pool_tasks:
                        if p["Pool Name"] == data:
                            res = p[textkey.split(".")[-1].title()]
                            break
                else:
                    for p in dropped_message:
                        if p["Message type"] == data:
                            res = p["Dropped"]
                            break
            else:
                cfstats = get_cfstats(config, keyspace=keyspace, table=table)
                if keyspace in cfstats:
                    if table:
                        return cfstats[keyspace]["tables"][table].get(textkey)
                    else:
                        return cfstats[keyspace].get(textkey)

# Shouldn't get here
        return None