MDS op times

mds-op-times.py

import re
import sys
import stat
import datetime

def parselog(logfile):
    client_request_map = {}
    with open(logfile, 'r') as logf:
        for loge in logf:
            is_incoming_request = True
            # try match request start
            m = re.match(r"(\d+\-\d+-\d+T\d+\:\d+\:\d+\.\d+)\+000.*handle_client_request client_request\(client\.(\d+)\:(\d+) .*", loge)
            if not m:
                # try match request reply
                m = re.match(r"(\d+\-\d+-\d+T\d+\:\d+\:\d+\.\d+)\+000.*reply_client_request .* client_request\(client\.(\d+)\:(\d+) .*", loge)
                if not m:
                    continue
                is_incoming_request = False
            dt = datetime.datetime.strptime(m.group(1), "%Y-%m-%dT%H:%M:%S.%f")
            dts = dt.timestamp()
            clientid = int(m.group(2))
            reqid = int(m.group(3))
            #print(f'{loge}')
            if is_incoming_request:
                # first time seen this client
                if not clientid in client_request_map:
                    client_request_map[clientid] = {}
                client_request_map[clientid][reqid] = [dts, dts]
            else:
                # log does not have request start
                if not clientid in client_request_map or \
                   not reqid in client_request_map[clientid]:
                    continue
                client_request_map[clientid][reqid][1] = dts
        for clientid,reqmap in client_request_map.items():
            sm = sorted(reqmap.items(), key=lambda val: val[1][1] - val[1][0], reverse=True)
            print(f'client.{clientid}')
            for e in sm[:5]:
                print(f'    requestid: {e[0]}, took {e[1][1] - e[1][0]}s')

if __name__ == '__main__':
    if len(sys.argv) < 2:
        print(f'usage: {sys.argv[0]} /path/to/mds/debug.log')
        sys.exit(-1)
    parselog(sys.argv[1])