Thread: [PERFORM] Unable to connect to server

Hi,

today i head a issue with pgsql 9.3, a lot of clients complained that they are unable to connect to the server.
when i have checked the logs i sow a lot of
FATAL:  canceling authentication due to timeout

server configuration is:
Ubuntu 16.04
128G RAM
RAID 10 4X 2T HDD
Intel(R) Xeon(R) CPU E5-1650 v2 @ 3.50GHz

clients are connecting remotely to the server, maybe i have misconfigured someting in postgresql setup.

pgsql config:
data_directory = '/var/lib/postgresql/9.3/main'         # use data in another directory
hba_file = '/etc/postgresql/9.3/main/pg_hba.conf'       # host-based authentication file
ident_file = '/etc/postgresql/9.3/main/pg_ident.conf'   # ident configuration file
external_pid_file = '/var/run/postgresql/9.3-main.pid'                  # write an extra PID file
listen_addresses = '*'          # what IP address(es) to listen on;
port = 5432                             # (change requires restart)
max_connections = 1000                  # (change requires restart)
unix_socket_directories = '/var/run/postgresql' # comma-separated list of directories
ssl = on                                # (change requires restart)
ssl_ciphers = 'DEFAULT:!LOW:!EXP:!MD5:@STRENGTH'        # allowed SSL ciphers
ssl_cert_file = 'server.crt'                # (change requires restart)
ssl_key_file = 'server.key'         # (change requires restart)
password_encryption = on
max_files_per_process = 5000            # min 25
                                        # (change requires restart)
vacuum_cost_delay =     20              # 0-100 milliseconds
vacuum_cost_page_hit = 1                # 0-10000 credits
vacuum_cost_page_miss = 10              # 0-10000 credits
vacuum_cost_page_dirty = 20             # 0-10000 credits
vacuum_cost_limit = 200         # 1-10000 credits
                                        # (change requires restart)
fsync = on                              # turns forced synchronization on or off
synchronous_commit = on                 # synchronization level;
                                        # off, local, remote_write, or on
wal_sync_method = fsync                 # the default is the first option

full_page_writes = on                   # recover from partial page writes
wal_buffers = -1                        # min 32kB, -1 sets based on shared_buffers
                                        # (change requires restart)
wal_writer_delay = 200ms                # 1-10000 milliseconds
commit_delay = 0                        # range 0-100000, in microseconds
commit_siblings = 5                     # range 1-1000
checkpoint_segments = 64                # in logfile segments, min 1, 16MB each
checkpoint_timeout = 15min              # range 30s-1h
checkpoint_completion_target = 0.9      # checkpoint target duration, 0.0 - 1.0
checkpoint_warning = 30s                # 0 disables
 
log_checkpoints = on
log_line_prefix = '%t [%p-%l] %q%u@%d '                 # special values:
 
track_activities = on
track_counts = on
stats_temp_directory = '/var/run/postgresql/9.3-main.pg_stat_tmp'
autovacuum = on                         # Enable autovacuum subprocess?  'on'
                                        # requires track_counts to also be on.
log_autovacuum_min_duration = 0         # -1 disables, 0 logs all actions and
autovacuum_max_workers = 5              # max number of autovacuum subprocesses
                                        # (change requires restart)
autovacuum_naptime = 5min               # time between autovacuum runs
autovacuum_vacuum_threshold = 500       # min number of row updates before
                                        # vacuum
autovacuum_analyze_threshold = 500      # min number of row updates before
                                        # analyze
autovacuum_vacuum_scale_factor = 0.4    # fraction of table size before vacuum
autovacuum_analyze_scale_factor = 0.2   # fraction of table size before analyze

autovacuum_vacuum_cost_delay = 20ms     # default vacuum cost delay for

datestyle = 'iso, mdy'
timezone = 'Europe/Bucharest'

client_encoding = sql_ascii             # actually, defaults to database
                                        # encoding
lc_messages = 'C'                       # locale for system error message
                                        # strings
lc_monetary = 'C'                       # locale for monetary formatting
lc_numeric = 'C'                        # locale for number formatting
lc_time = 'C'                           # locale for time formatting
default_text_search_config = 'pg_catalog.english'

default_statistics_target = 100 # pgtune wizard 2016-12-11
maintenance_work_mem = 1GB # pgtune wizard 2016-12-11
constraint_exclusion = on # pgtune wizard 2016-12-11
effective_cache_size = 88GB # pgtune wizard 2016-12-11
work_mem = 64MB # pgtune wizard 2016-12-11
wal_buffers = 32MB # pgtune wizard 2016-12-11
shared_buffers = 30GB # pgtune wizard 2016-12-11

Thanks,

HI Andreas,

thanks for your replay, i have checked connection number for the last 3 mouths and maximum number of connection was 300, so i can reduce this is it's cousin the issue.
during this error server was 99% idle, the only thing is that there was a pg_dump running on one of the DB, can this cause this error, also i have restarted pg server 2 days before because of server cert update.

---